[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-mkiol--dsnote":3,"tool-mkiol--dsnote":64},[4,17,25,39,48,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,14,15],"开发框架","Agent","语言模型","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":10,"last_commit_at":23,"category_tags":24,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,15],{"id":26,"name":27,"github_repo":28,"description_zh":29,"stars":30,"difficulty_score":10,"last_commit_at":31,"category_tags":32,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[33,34,35,36,14,37,15,13,38],"图像","数据工具","视频","插件","其他","音频",{"id":40,"name":41,"github_repo":42,"description_zh":43,"stars":44,"difficulty_score":45,"last_commit_at":46,"category_tags":47,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,3,"2026-04-04T04:44:48",[14,33,13,15,37],{"id":49,"name":50,"github_repo":51,"description_zh":52,"stars":53,"difficulty_score":45,"last_commit_at":54,"category_tags":55,"status":16},519,"PaddleOCR","PaddlePaddle\u002FPaddleOCR","PaddleOCR 是一款基于百度飞桨框架开发的高性能开源光学字符识别工具包。它的核心能力是将图片、PDF 等文档中的文字提取出来，转换成计算机可读取的结构化数据，让机器真正“看懂”图文内容。\n\n面对海量纸质或电子文档，PaddleOCR 解决了人工录入效率低、数字化成本高的问题。尤其在人工智能领域，它扮演着连接图像与大型语言模型（LLM）的桥梁角色，能将视觉信息直接转化为文本输入，助力智能问答、文档分析等应用场景落地。\n\nPaddleOCR 适合开发者、算法研究人员以及有文档自动化需求的普通用户。其技术优势十分明显：不仅支持全球 100 多种语言的识别，还能在 Windows、Linux、macOS 等多个系统上运行，并灵活适配 CPU、GPU、NPU 等各类硬件。作为一个轻量级且社区活跃的开源项目，PaddleOCR 既能满足快速集成的需求，也能支撑前沿的视觉语言研究，是处理文字识别任务的理想选择。",74913,"2026-04-05T10:44:17",[15,33,13,37],{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":45,"last_commit_at":62,"category_tags":63,"status":16},2181,"OpenHands","OpenHands\u002FOpenHands","OpenHands 是一个专注于 AI 驱动开发的开源平台，旨在让智能体（Agent）像人类开发者一样理解、编写和调试代码。它解决了传统编程中重复性劳动多、环境配置复杂以及人机协作效率低等痛点，通过自动化流程显著提升开发速度。\n\n无论是希望提升编码效率的软件工程师、探索智能体技术的研究人员，还是需要快速原型验证的技术团队，都能从中受益。OpenHands 提供了灵活多样的使用方式：既可以通过命令行（CLI）或本地图形界面在个人电脑上轻松上手，体验类似 Devin 的流畅交互；也能利用其强大的 Python SDK 自定义智能体逻辑，甚至在云端大规模部署上千个智能体并行工作。\n\n其核心技术亮点在于模块化的软件智能体 SDK，这不仅构成了平台的引擎，还支持高度可组合的开发模式。此外，OpenHands 在 SWE-bench 基准测试中取得了 77.6% 的优异成绩，证明了其解决真实世界软件工程问题的能力。平台还具备完善的企业级功能，支持与 Slack、Jira 等工具集成，并提供细粒度的权限管理，适合从个人开发者到大型企业的各类用户场景。",70612,"2026-04-05T11:12:22",[15,14,13,36],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":75,"owner_avatar_url":76,"owner_bio":77,"owner_company":77,"owner_location":77,"owner_email":77,"owner_twitter":77,"owner_website":77,"owner_url":78,"languages":79,"stars":100,"forks":101,"last_commit_at":102,"license":103,"difficulty_score":10,"env_os":104,"env_gpu":105,"env_ram":106,"env_deps":107,"category_tags":118,"github_topics":119,"view_count":10,"oss_zip_url":77,"oss_zip_packed_at":77,"status":16,"created_at":135,"updated_at":136,"faqs":137,"releases":171},3777,"mkiol\u002Fdsnote","dsnote","Speech Note Linux app. Note taking, reading and translating with offline Speech to Text, Text to Speech and Machine translation.","dsnote 是一款专为 Linux 桌面和 Sailfish OS 设计的智能笔记应用，集语音转文字、文字转语音及机器翻译功能于一体。它致力于解决用户在记录、阅读和跨语言交流时的效率痛点，让用户能够轻松通过语音输入笔记、聆听文档内容或实时翻译文本。\n\n这款工具特别适合注重隐私保护的普通用户、需要离线办公的专业人士以及多语言学习者使用。其最核心的技术亮点在于“完全离线”运行：所有的语音识别、合成与翻译过程均在本地计算机完成，无需连接互联网，确保您的数据永远不会上传至云端，彻底杜绝隐私泄露风险。\n\ndsnote 并非依赖单一引擎，而是灵活集成了多种先进的开源模型，包括 Whisper、Vosk、Faster Whisper 用于语音识别，Piper、Coqui TTS 等用于语音合成，以及 Bergamot 用于翻译。用户可通过应用内置的模型浏览器，按需下载并切换支持数十种语言的模型组合。无论是希望在不联网环境下高效工作的用户，还是想体验最新本地化 AI 技术的爱好者，dsnote 都是一个安全、自由且强大的选择。","# Speech Note\n\nLinux desktop and Sailfish OS app for note taking, reading and translating with offline Speech to Text, Text to Speech and Machine Translation\n\n\u003Ca href='https:\u002F\u002Fflathub.org\u002Fapps\u002Fnet.mkiol.SpeechNote'>\u003Cimg width='240' alt='Download on Flathub' src='https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmkiol_dsnote_readme_410555b9dfc2.png'\u002F>\u003C\u002Fa>\n\n## Contents of this README\n\n- [Description](#description)\n- [Languages and Models](#languages-and-models)\n- [How to install](#how-to-install)\n- [Flatpak packages](#flatpak-packages)\n- [Beta version](#beta-version)\n- [Extra features](#extra-features)\n- [Building from sources](#building-from-sources)\n- [How to enable a custom model](#how-to-enable-a-custom-model)\n- [Contributing to Speech Note](#contributing-to-speech-note)\n- [How to support](#how-to-support)\n- [Reviews and demos](#reviews-and-demos)\n- [License](#license)\n\n## Description\n\n**Speech Note** let you take, read and translate notes in multiple languages.\nIt uses Speech to Text, Text to Speech and Machine Translation to do so.\nText and voice processing take place entirely offline, locally on your\ncomputer, without using a network connection. Your privacy is always\nrespected. No data is sent to the Internet.\n\n**Speech Note** uses many different processing engines to do its job.\nCurrently these are used:\n\n- Speech to Text (STT)\n - [Coqui STT (a fork of Mozilla DeepSpeech)](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FSTT)\n - [Vosk](https:\u002F\u002Falphacephei.com\u002Fvosk)\n - [whisper.cpp](https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fwhisper.cpp)\n - [Faster Whisper](https:\u002F\u002Fgithub.com\u002Fguillaumekln\u002Ffaster-whisper)\n - [april-asr](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr)\n- Text to Speech (TTS)\n - [espeak-ng](https:\u002F\u002Fgithub.com\u002Fespeak-ng\u002Fespeak-ng)\n - [MBROLA](https:\u002F\u002Fgithub.com\u002Fnumediart\u002FMBROLA)\n - [Piper](https:\u002F\u002Fgithub.com\u002Frhasspy\u002Fpiper)\n - [RHVoice](https:\u002F\u002Fgithub.com\u002FRHVoice\u002FRHVoice)\n - [Coqui TTS](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FTTS)\n - [Mimic 3](https:\u002F\u002Fmycroft.ai\u002Fmimic-3)\n - [WhisperSpeech](https:\u002F\u002Fcollabora.github.io\u002FWhisperSpeech\u002F)\n - [Kokoro](https:\u002F\u002Fgithub.com\u002Fhexgrad\u002Fkokoro)\n - [Parler-TTS](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fparler-tts)\n - [F5-TTS](https:\u002F\u002Fgithub.com\u002FSWivid\u002FF5-TTS)\n - [S.A.M.](https:\u002F\u002Fdiscordier.github.io\u002Fsam)\n- Machine Translation (MT)\n - [Bergamot Translator](https:\u002F\u002Fgithub.com\u002Fbrowsermt\u002Fbergamot-translator)\n\n## Languages and Models\n\nSpeech Note installation package does not include checkpoint files for supported models, but instead they can be easily downloaded using the graphical model browser built into the application.\n\nFollowing languages and models are supported and enable for download:\n\n| **Lang ID** | **Name** | **DeepSpeech (STT)** | **Whisper (STT)** | **Vosk (STT)** | **April-ASR (STT)** | **Piper (TTS)** | **RHVoice (TTS)** | **espeak (TTS)** | **MBROLA (TTS)** | **Coqui (TTS)** | **Mimic3 (TTS)** | **WhisperSpeech (TTS)** | **Kokoro (TTS)** | **F5-TTS** | **Parler-TTS** | **S.A.M. (TTS)** | **Bergamot (MT)** |\n| ----------- | ------------- | -------------------- | ----------------- | -------------- | ------------------- | --------------- | ----------------- | ---------------- | ---------------- | --------------- | ---------------- | ----------------------- | ---------------- | ---------- | -------------- | ---------------- | ----------------- |\n| af | Afrikaans | | ● | | | | | ● | | | ● | | | | | | |\n| am | Amharic | ● (e) | ● | | | | | ● | | ● | | | | | | | |\n| ar | Arabic | | ● | ● | | ● | | ● | ● | ● | | | | | | | ● |\n| az | Azerbaijani | | ● | | | | | | | | | | | | | | ● |\n| be | Belarusian | | ● | | | | | | | | | | | | | | ● |\n| bg | Bulgarian | | ● | | | | | ● | | ● | | | | | | | |\n| bn | Bengali | | ● | | | | | ● | | ● | ● | | | | | | |\n| bs | Bosnian | | ● | | | | | ● | | | | | | | | | ● |\n| ca | Catalan | ● | ● | ● | | ● | | ● | | ● | | | | | | | ● |\n| cs | Czech | ● | ● | ● | | ● | ● | ● | ● | ● | | | | | | | ● |\n| cy | Welsh | | | | | ● | | | | | | | | | | | |\n| da | Danish | | ● | | | ● | | ● | | ● | | | | | | | ● |\n| de | German | ● | ● | ● | | ● | | ● | | ● | ● | ● | | | ●(e) | | ● |\n| el | Greek | ● (e) | ● | | | ● | | ● | | ● | ● | | | | | | ● |\n| en | English | ● | ● | ● | ● | ● | ● | ● | | ● | ● | ● | ● | ● | ● | ● | ● |\n| eo | Esperanto | | | ● | | | ● | ● | | | | | | | | | |\n| es | Spanish | ● | ● | ● | | ● | ● | ● | | ● | ● | ● | ● | | ●(e) | | ● |\n| et | Estonian | ● (e) | ● | | | | | ● | ● | ● | | | | | | | ● |\n| eu | Basque | ● (e) | ● | | | | | ● | | ● | | | | | | | |\n| fa | Persian | ● | ● | ● | | ● | | ● | ● | ● | ● | | | | | | ● |\n| fi | Finnish | ● | ● | | | ● | | ● | | ● | ● | | | | | | ● |\n| fr | French | ● | ● | ● | ● | ● | | ● | | ● | ● | ● | ● | | ●(e) | | ● |\n| ga | Irish | | | | | | | ● | | ● | | | | | | | |\n| gu | Gujarati | | ● | | | | | ● | | | ● | | | | | | |\n| ha | Hausa | | ● | | | | | | | | ● | | | | | | |\n| he | Hebrew | | ● | | | | | | | ● | | | | | | | |\n| hi | Hindi | | ● | ● | | ● | | ● | | | | | ● | | | | |\n| hr | Croatian | | ● | | | | ● | ● | ● | ● | | | | | | | |\n| hu | Hungarian | ● (e) | ● | | | ● | | ● | ● | ● | ● | | | | | | ● |\n| id | Indonesian | ● (e) | ● | | | | | ● | ● | ● | | | | | | | ● |\n| is | Icelandic | | ● | | | ● | | ● | | ● | | | | | | | ● |\n| it | Italian | ● | ● | ● | | ● | | ● | | ● | ● | ● | ● | | ●(e) | | ● |\n| ja | Japanese | | ● | ● | | | | ● | | ● | | | ● | | | | ● |\n| jv | Javanese | | ● | | | | | | | | ● | | | | | | |\n| ka | Georgian | | ● | | | ● | ● | ● | | | | | | | | | |\n| kk | Kazakh | | ● | ● | | ● | | ● | | ● | | | | | | | |\n| kn | Kannada | | ● | | | | | | | ● | | | | | | | ● |\n| ko | Korean | | ● | ● | | | | ● | | ● | | | | | | | ● |\n| ky | Kyrgyz | | | | | | ● | ● | | | | | | | | | |\n| la | Latin | | | | | | | ● | | ● | | | | | | | |\n| lb | Luxembourgish | | | | | ● | | | | | | | | | | | |\n| lt | Lithuanian | | ● | | | | | ● | ● | ● | | | | | | | ● |\n| lv | Latvian | ● | ● | | | ● | | ● | | ● | | | | | | | ● |\n| mk | Macedonian | | ● | | | | ● | ● | | | | | | | | | |\n| ml | Malayalam | | ● (e) | | | ● | | | | ● | | | | | | | ● |\n| mn | Mongolian | ● (e) | ● | | | | | | | ● | | | | | | | |\n| mr | Marathi | | ● | | | | | | | ● | | | | | | | |\n| ms | Malay | | ● | | | | | ● | ● | ● | | | | | | | |\n| mt | Maltese | | ● | | | | | ● | | ● | | | | | | | |\n| ne | Nepali | | ● | | | ● | | ● | | | ● | | | | | | |\n| nl | Dutch | ● (e) | ● | ● | | ● | | ● | | ● | ● | ● | | | ●(e) | | ● |\n| no | Norwegian | | ● | | | ● | | ● | | | | | | | | | ● |\n| pl | Polish | ● | ● | ● | ● | ● | ● | ● | ● | ● | ● | ● | | | ●(e) | | ● |\n| pt | Portuguese | ● (e) | ● | ● | | ● | | ● | ● | ● | | | ● | | ●(e) | | ● |\n| ro | Romanian | ● (e) | ● | | | ● | | ● | ● | ● | | | | | | | ● |\n| ru | Russian | ● | ● | ● | | ● | ● | ● | | | ● | | | | | | ● |\n| sk | Slovak | | ● | | | ● | ● | ● | | ● | | | | | | | ● |\n| sl | Slovenian | ● (e) | ● | | | ● | | ● | | ● | | | | | | | ● |\n| sq | Albanian | | ● | | | | ● | ● | | ● | | | | | | | |\n| sr | Serbian | | ● | | | ● | ● | ● | | | | | | | | | ● |\n| sv | Swedish | | ● | ● | | ● | | ● | ● | ● | ● | | | | | | ● |\n| sw | Swahili | ● | ● | | | ● | | ● | | ● | | | | | | | |\n| ta | Tamil | | ● | | | | | | | ● | | | | | | | ● |\n| te | Telugu | | ● | | | | | ● | | | ● | | | | | | |\n| th | Thai | ● (e) | ● | | | | | ● | | ● | | | | | | | |\n| tl | Tagalog | | ● | ● | | | | | | ● | | | | | | | |\n| tn | Tswana | | ● | | | | | ● | | | ● | | | | | | |\n| tr | Turkish | ● (e) | ● | ● | | ● | | ● | ● | ● | | | | | | | ● |\n| tt | Tatar | | ● | | | | ● | ● | | ● | | | | | | | |\n| uk | Ukrainian | ● | ● | ● | | ● | ● | ● | | ● | ● | | | | | | ● |\n| uz | Uzbek | | ● | ● | | | | ● | | ● | | | | | | | |\n| vi | Vietnamese | | ● | ● | | ● | | ● | | ● | | | | | | | ● |\n| yo | Yoruba | ● (e) | ● | | | | | | | ● | ● | | | | | | |\n| zh | Chinese | ● | ● | ● | | ● | | ● | | ● | | | ● | ● | | | ● |\n\n\u003Csup>(e) experimental, most likely doesn't work well\u003C\u002Fsup>\n\u003Cbr\u002F>\n\nFaster Whisper, Coqui TTS and Mimic3 models are only available on x86-64.\n\nLanguage models can be downloaded directly from the app.\n\nDetails of models which are currently configured for download are described in\n[models.json (GitHub)](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Fblob\u002Fmain\u002Fconfig\u002Fmodels.json) or\n[models.json (GitLab)](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote\u002F-\u002Fblob\u002Fmain\u002Fconfig\u002Fmodels.json).\n\n## How to install\n\n- Linux Desktop: [Flatpak](https:\u002F\u002Fflathub.org\u002Fapps\u002Fnet.mkiol.SpeechNote)\n\n```sh\n# Flatpak base package\nflatpak install net.mkiol.SpeechNote\n\n# Optional NVIDIA add-on package\nflatpak install net.mkiol.SpeechNote.Addon.nvidia\n\n# Optional AMD add-on package\nflatpak install net.mkiol.SpeechNote.Addon.amd\n```\n\n- Arch Linux (AUR):\n\n - [dsnote](https:\u002F\u002Faur.archlinux.org\u002Fpackages\u002Fdsnote)\n - [dsnote-git](https:\u002F\u002Faur.archlinux.org\u002Fpackages\u002Fdsnote-git)\n\n- openSUSE ([Packman repository](https:\u002F\u002Fen.opensuse.org\u002FAdditional_package_repositories#Packman))\n\n```sh\n# Base package\nzypper in speechnote\n\n# Optional support for Python-based features in Speech Note\nzypper in speechnote-python-modules\n```\n\n- Sailfish OS: [OpenRepos](https:\u002F\u002Fopenrepos.net\u002Fcontent\u002Fmkiol\u002Fspeech-note)\n\n### Flatpak packages\n\nThe app distributed via Flatpak (published on Flathub) consists of the following packages:\n\n- Base package \"Speech Note\" (net.mkiol.SpeechNote)\n- Optional add-on for NVIDIA graphics card \"Speech Note NVIDIA\" (net.mkiol.SpeechNote.Addon.nvidia)\n- Optional add-on for AMD graphics card \"Speech Note AMD\" (net.mkiol.SpeechNote.Addon.amd)\n\nBase package includes all the dependencies needed to run every feature of the application.\nAdd-ons add the capability of GPU acceleration, which speeds up some operations in the application.\n\nBase package and add-ons contain many \"heavy\" libraries like CUDA, ROCm, Torch and Python libraries.\nDue to this, the size of the packages and the space required after installation are significant.\nIf you don't need all the functionalities, you can use much smaller \"Tiny\" package\n(available on [Releases](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Freleases) page),\nwhich provides only the basic features. If you need, you can also use \"Tiny\" packages together with GPU acceleration add-on.\n\nSize comparison between Base, Tiny and Add-ons Flatpak packages:\n\n| **Sizes** | **Base** | **Tiny** | **AMD add-on** | **NVIDIA add-on** |\n| ------------- | ---------| ---------| -------------- | ----------------- |\n| Download size | 1.2 GiB | 48 MiB | +7.6 GiB | +4.3 GiB |\n| Unpacked size | 3.6 GiB | 170 MiB | +34 GiB | +9 GiB |\n\nAdd-on packages use so-called \"extra-data\". This means that during installation, several binary modules are downloaded from the original sources and merged with the Flatpak package. Due to this process, additional disk space is temporarily required when installing the add-on. Make sure you have enough disk space on the partition where flatpak applications are installed. When the Flatpak package is installed system-wide (default), the applications are stored in _\u002Fvar\u002Flib\u002Fflatpak_. In case of user installation (--user), the apps are stored in _~\u002F.local\u002Fshare\u002Fflatpak_.\n\nDisk space required for the installation of the add-on package:\n\n| **Add-on** | **Size** |\n| --------------| ---------|\n| AMD add-on | 55 GiB |\n| NVIDIA add-on | 15 GiB |\n\nFeatures comparison between Base, Tiny and Add-ons Flatpak packages:\n\n| **Features** | **Base** | **Tiny** | **AMD add-on** | **NVIDIA add-on** |\n| --------------------------------------- | ---------| ---------| -------------- | ----------------- |\n| Coqui\u002FDeepSpeech STT | + | + | | |\n| Vosk STT | + | + | | |\n| Whisper (whisper.cpp) STT | + | + | | |\n| Whisper (whisper.cpp) STT OpenCL ROCm | - | - | + | |\n| Whisper (whisper.cpp) STT OpenCL NVIDIA | + | + | | |\n| Whisper (whisper.cpp) STT ROCm | - | - | + | |\n| Whisper (whisper.cpp) STT CUDA | - | - | | + |\n| Whisper (whisper.cpp) STT OpenVINO | + | - | | |\n| Whisper (whisper.cpp) STT Vulkan | + | + | | |\n| FasterWhisper STT | + | - | | |\n| FasterWhisper STT CUDA | - | - | | + |\n| April-ASR STT | + | + | | |\n| eSpeak TTS | + | + | | |\n| MBROLA TTS | + | + | | |\n| Piper TTS | + | + | | |\n| RHVoice TTS | + | + | | |\n| Coqui TTS | + | - | | |\n| Coqui TTS ROCm | - | - | + | |\n| Coqui TTS CUDA | - | - | | + |\n| Mimic3 TTS | + | - | | |\n| WhisperSpeech TTS | + | - | | |\n| WhisperSpeech TTS ROCm | - | - | + | |\n| WhisperSpeech TTS CUDA | - | - | | + |\n| Kokoro TTS | + | - | | |\n| Kokoro TTS ROCm | - | - | + | |\n| Kokoro TTS CUDA | - | - | | + |\n| Parler-TTS | + | - | | |\n| Parler-TTS ROCm | - | - | + | |\n| Parler-TTS CUDA | - | - | | + |\n| F5-TTS | + | - | | |\n| F5-TTS ROCm | - | - | + | |\n| F5-TTS CUDA | - | - | | + |\n| S.A.M TTS | + | + | | |\n| Punctuation restoration | + | - | | |\n| Translator | + | + | | |\n\n### Beta version\n\nIn addition to the stable version in the Flathub repository, you can try to test the \"Beta\" version of the upcoming release. This version is usable, but may contain more bugs.\n\nBeta version is available in \"flathub-beta\" repository. Follow [these instructions](https:\u002F\u002Fdiscourse.flathub.org\u002Ft\u002Fhow-to-use-flathub-beta\u002F2111) to enable flathub-beta on your computer.\n\n## Extra features\n\n### Command-line options\n\nThe CLI interface is primary for the integration with the desktop when _Speech Note_ is already running (for example, hidden in the system tray or in the backgraund).\n\n#### Examples\n\nList all supported options:\n\n```shell\nflatpak run net.mkiol.SpeechNote --help\n```\n\nStart listening:\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-listening\n```\n\nCancel any already started action:\n\n```sh\nflatpak run net.mkiol.SpeechNote --action cancel\n```\n\nStart listening, the decoded text will be saved to the clipboard:\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-listening-clipboard\n```\n\nStart listening, the decoded text will be inserted into any window on the desktop on which the cursor is focused:\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-listening-active-window\n```\n\nStart reading \"Hello, how are you doing?\":\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-reading-text --text \"Hello, how are you doing?\"\n```\n\nSave speech of \"Hello, how are you doing?\" to \"speech.mp3\" file:\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-reading-text --text \"Hello, how are you doing?\" --output-file speech.mp3\n```\n\nList all available TTS models:\n\n```sh\nflatpak run net.mkiol.SpeechNote --print-available-models tts\n```\n\n### Global keyboard shortcuts\n\nGlobal keyboard shortcuts allow you to start listening or reading with the keyboard, even when the application is not active (e.g. minimized, hidden in the system tray icon or just in the background).\n\nTo enable and customize keyboard shortcuts, go to `Settings`->`Accessibility`->`Use Global Keyboard Shortcuts`.\n\nIn order for shortcuts to work under Wayland, your desktop environment must support [GlobalShortcuts interface](https:\u002F\u002Fflatpak.github.io\u002Fxdg-desktop-portal\u002Fdocs\u002Fdoc-org.freedesktop.impl.portal.GlobalShortcuts.html) in the _XDG Desktop Portal_ service. Currently, `GlobalShortcuts` is supported only in the latest KDE Plasma and GNOME desktops.\n\nWhen `XDG Desktop Portal` is used to manage global shortcuts, use the desktop environment tool to customize key bindings.\n\n### Insert into active window\n\nUsing global keyboard shortcuts or command-line actions, you can directly start listening and insert the decoded text into any window that is currently in focus. This allows you to use _Speech Note_ as a voice typing tool on the desktop.\n\nUnder X11, this feature should work right out of the box.\n\nUnder Wayland, the external `ydotool` daemon must be installed and running for it to work. If you are using Flatpak, also make sure that the application has permission to access the `ydotool` daemon's socket file and the socket file.\n\n### Subtitles\n\nWhen transcribing (STT), you can also convert audio to subtitles instead of text. The supported format is SRT.\n\nSubtitles are also supported in TTS. The reading aloud of subtitles is synchronized with the timestamps of the subtitles. If configured, the speed of the voice can be automatically adjusted to the duration of the subtitle segment. This allows you to create a voice-over from the subtitle file.\n\n## Building from sources\n\n### Arch Linux\n\nIt is also possible to build and install the latest development (git) or latest stable (release) version from the repository using the provided PKGBUILD file (please note that the same remarks about building on Linux apply):\n\n```sh\ngit clone \u003Cgit repository url>\n\ncd dsnote\u002Farch\u002Fgit # build latest git version\n# or\ncd dsnote\u002Farch\u002Frelease # build latest release version\n\nmakepkg -si\n```\n\n### RHEL\u002FFedora\u002FRocky Linux\n\nIt is also possible to build and install the latest development version from the repository using the provided SPEC file and helper `make_rpm.sh` script:\n\n```sh\ngit clone \u003Cgit repository url>\n\ncd dsnote\u002Ffedora\n\n# optionally install build dependencies\ndnf install rpmdevtools autoconf automake boost-devel cmake git kf5-kdbusaddons-devel libarchive-devel libxdo-devel libXinerama-devel libxkbcommon-x11-devel libXtst-devel libtool meson openblas-devel patchelf pybind11-devel python3-devel python3-pybind11 qt5-linguist qt5-qtmultimedia-devel qt5-qtquickcontrols2-devel qt5-qtx11extras-devel rubberband-devel taglib-devel vulkan-headers\n\n.\u002Fmake_rpm.sh\n```\n\n### Flatpak\n\n```sh\ngit clone \u003Cgit repository url>\n\ncd dsnote\u002Fflatpak\n\n# build a base package\nflatpak-builder --force-clean --user --install-deps-from=flathub --repo=\"\u003Cname or \u002Fpath\u002Fto\u002Flocal\u002Fflatpak\u002Frepo>\" \"\u002Fpath\u002Fto\u002Foutput\u002Fdir\" net.mkiol.SpeechNote.yaml\n\n# build an optional NVIDIA add-on package\nflatpak-builder --force-clean --user --install-deps-from=flathub --repo=\"\u003Cname or \u002Fpath\u002Fto\u002Flocal\u002Fflatpak\u002Frepo>\" \"\u002Fpath\u002Fto\u002Foutput\u002Fdir\" net.mkiol.SpeechNote.Addon.nvidia.yaml\n```\n\n### Sailfish OS\n\n```sh\ngit clone \u003Cgit repository url>\n\ncd dsnote\nmkdir build\ncd build\n\nsfdk config --session specfile=..\u002Fsfos\u002Fharbour-dsnote.spec\nsfdk config --session target=SailfishOS-4.4.0.58-aarch64\nsfdk cmake ..\u002F -DCMAKE_BUILD_TYPE=Release -DWITH_SFOS=ON -DWITH_PY=OFF\nsfdk package\n```\n\n### Linux (direct build)\n\nSpeech Note has many build-time and run-time dependencies. This includes shared and static libraries,\n3rd-party executables, Python and Perl scripts. Because of these complexity, the recommended way to build\nis to use Flatpak tool-chain (Flatpak manifest file and [flatpak-builder](https:\u002F\u002Fdocs.flatpak.org\u002Fen\u002Flatest\u002Fflatpak-builder.html)).\nIf you want to make a direct build (i.e. without flatpak) it is also possible but more complicated.\n\n```sh\ngit clone \u003Cgit repository url>\n\ncd dsnote\nmkdir build\ncd build\n\ncmake ..\u002F -DCMAKE_BUILD_TYPE=Release -DWITH_DESKTOP=ON\nmake\n```\n\nTo make build without support for Python components, add `-DWITH_PY=OFF` in cmake step.\n\nTo see other build options search for `option(BUILD_XXX)` in `CMakeList.txt` file.\n\n### Minimal Linux (direct build)\n\nFor developing features locally, going through the whole build process can be time-consuming; use this to make a minimal direct build.\n\n```sh\ngit clone \u003Cgit repository url>\n\ncd dsnote\nmkdir build\ncd build\n\ncmake ..\u002F -DWITH_DESKTOP=ON \\\n -DWITH_PY=ON \\\n -DDOWNLOAD_VOSK=ON \\\n -DBUILD_VOSK=OFF \\\n -DBUILD_WHISPERCPP=OFF \\\n -DBUILD_OPENBLAS=OFF \\\n -DBUILD_RHVOICE=OFF \\\n -DBUILD_RHVOICE_MODULE=OFF \\\n -DBUILD_BERGAMOT=OFF\nmake\n```\n\n## Building WlClipboard\n\nIf you want to build wl-clipboard or if you do not have access to wl-clipboard in your system, and KDE Klipper or QClipboard are not working when you use simulated Ctrl+V when pasting text into the active window.\n\nYou can add this flag to build wl-clipboard:\n\n```sh\n\ncmake ..\u002F -DWITH_DESKTOP=ON -DCMAKE_BUILD_TYPE=Release -DBUILD_WL_CLIPBOARD\n```\n\n## How to enable a custom model\n\nAll models available for download are specified in the configuration file (config\u002Fmodels.json).\nTo enable a custom model that is compatible with currently supported engines, simply edit this file and restart the application.\n\nWhen you first run the application, the models configuration file is created in:\n\n- `~\u002F.local\u002Fshare\u002Fnet.mkiol\u002Fdsnote\u002Fmodels.json`, or\n- `~\u002F.var\u002Fapp\u002Fnet.mkiol.SpeechNote\u002Fdata\u002Fnet.mkiol\u002Fdsnote\u002Fmodels.json` (Flatpak), or\n- `~\u002F.local\u002Fshare\u002Forg.mkiol\u002Fdsnote\u002Fmodels.json` (Sailfish OS)\n\nYou can freely edit currently enabled models or add new ones.\n\nModel definition looks like this:\n\n```json\n{\n \"name\": \"\u003Cmodel name>\",\n \"model_id\": \"\u003Cmodel unique id>\",\n \"engine\": \"\u003Cengine type>\",\n \"lang_id\": \"\u003Clang id>\",\n \"checksum\": \"\u003Cmd5 checksum>\",\n \"checksum_quick\": \"\u003Cpartial md5 checksum>\",\n \"comp\": \"\u003Ccompression type\",\n \"urls\": [\n \u003Cmodel URLs>\n ],\n \"size\": \"\u003Cdownload size of all files>\"\n}\n```\n\nAllowed engine types: `stt_ds`, `stt_vosk`, `stt_april`, `stt_whisper`, `stt_fasterwhisper`, `tts_piper`, `tts_rhvoice`, `tts_espeak`, `tts_coqui`, `tts_mimic3`, `tts_whisperspeech`, `tts_sam`, `tts_parler`, `tts_f5`, `tts_kokoro`, `mnt_bergamot`\n\nAllowed compression types: `none`, `gz`, `xz`, `tarxz`, `targz`, `zip`, `zipall`, `dir`, `dirgz`\n\nAllowed URL types: `http`, `https`, `file`\n\nChecksums are calculated for all files after unpacking. If you are adding a new model, you can use the `--gen-checksums` command line option to find the right checksums. To do this, put empty strings in both `checksum` and `checksum_quick`, save the file and run Speech Note with the mentioned option.\n\nFor example:\n\n```json\n{\n \"name\": \"New Piper Voice\",\n \"model_id\": \"en_piper_new\",\n \"engine\": \"tts_piper\",\n \"lang_id\": \"en\",\n \"checksum\": \"\",\n \"checksum_quick\": \"\",\n \"size\": \"\"\n \"comp\": \"dir\",\n \"urls\": [\n \"file:\u002F\u002F\u002Fhome\u002Fme\u002Fmodels\u002Fnew-model-medium.onnx\",\n \"file:\u002F\u002F\u002Fhome\u002Fme\u002Fmodels\u002Fnew-model-medium.onnx.json\"\n ]\n}\n```\n\n```sh\nflatpak run net.mkiol.SpeechNote --verbose --gen-checksums\n```\n\n## Contributing to Speech Note\n\nAny contribution is very welcome!\n\nFor detailed information on how to contribute, please see [CONTRIBUTING.md](CONTRIBUTING.md).\n\nProject is hosted both on [GitHub](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote) and [GitLab](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote).\nFeel free to make a PR\u002FMR, report an issue or reqest for new feature on the platform you prefer the most.\n\n### Translation\n\nTranslation files in Qt format are in `translations` directory.\n\nPreferred way to contribute translation is via [Transifex service](https:\u002F\u002Fexplore.transifex.com\u002Fmkiol\u002Fdsnote\u002F),\nbut if you would like to make a direct PR\u002FMR, please do it.\n\n## How to support\n\nIf you find **Speech Note** useful and would like to support this project,\nplease consider doing one or two of the following:\n\n- Give a ⭐ on [GitHub](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote) or\u002Fand [GitLab](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote).\n- Write a review in your applications manager app (Discover, Software or any other).\n- Tell others about this app by mentioning it on social media.\n- If you have spare money, make a small donation via [ko-fi (one time)](https:\u002F\u002Fko-fi.com\u002Fmkiol) or [Liberapay (recurring)](https:\u002F\u002Fliberapay.com\u002Fmkiol\u002Fdonate).\n\n## Libraries\n\n**Speech Note** relies on following open source projects:\n\n- [Qt](https:\u002F\u002Fwww.qt.io\u002F)\n- [Coqui STT](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FSTT)\n- [Coqui TTS](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FTTS)\n- [Vosk](https:\u002F\u002Falphacephei.com\u002Fvosk)\n- [whisper.cpp](https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fwhisper.cpp)\n- [WebRTC VAD](https:\u002F\u002Fwebrtc.org\u002F)\n- [libarchive](https:\u002F\u002Flibarchive.org\u002F)\n- [RNNoise-nu](https:\u002F\u002Fgithub.com\u002FGregorR\u002Frnnoise-nu)\n- [{fmt}](https:\u002F\u002Ffmt.dev)\n- [Hugging Face Transformers](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Ftransformers)\n- [Piper](https:\u002F\u002Fgithub.com\u002Frhasspy\u002Fpiper)\n- [RHVoice](https:\u002F\u002Fgithub.com\u002FRHVoice\u002FRHVoice)\n- [ssplit-cpp](https:\u002F\u002Fgithub.com\u002Fugermann\u002Fssplit-cpp)\n- [espeak-ng](https:\u002F\u002Fgithub.com\u002Fespeak-ng\u002Fespeak-ng)\n- [bergamot-translator](https:\u002F\u002Fgithub.com\u002Fbrowsermt\u002Fbergamot-translator)\n- [Rubber Band Library](https:\u002F\u002Fbreakfastquay.com\u002Frubberband)\n- [simdjson](https:\u002F\u002Fsimdjson.org\u002F)\n- [Nlohmann JSON](https:\u002F\u002Fjson.nlohmann.me\u002F)\n- [uroman](https:\u002F\u002Fgithub.com\u002Fisi-nlp\u002Furoman)\n- [astrunc](https:\u002F\u002Fgithub.com\u002FJoke-Shi\u002Fastrunc)\n- [FFmpeg](https:\u002F\u002Fffmpeg.org\u002F)\n- [LAME](https:\u002F\u002Flame.sourceforge.io\u002F)\n- [Vorbis](https:\u002F\u002Fxiph.org\u002Fvorbis\u002F)\n- [TagLib](https:\u002F\u002Ftaglib.org\u002F)\n- [libnumbertext](https:\u002F\u002Fgithub.com\u002FNumbertext\u002Flibnumbertext)\n- [KDBusAddons](https:\u002F\u002Finvent.kde.org\u002Fframeworks\u002Fkdbusaddons)\n- [QHotkey](https:\u002F\u002Fgithub.com\u002FSkycoder42\u002FQHotkey)\n- [faster-whisper](https:\u002F\u002Fgithub.com\u002Fguillaumekln\u002Ffaster-whisper)\n- [Mimic 3](https:\u002F\u002Fmycroft.ai\u002Fmimic-3)\n- [Unikud](https:\u002F\u002Fgithub.com\u002Fmorrisalp\u002Funikud)\n- [april-asr](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr)\n- [Opus](https:\u002F\u002Fopus-codec.org\u002F)\n- [html2md](https:\u002F\u002Ftim-gromeyer.github.io\u002Fhtml2md\u002F)\n- [maddy](https:\u002F\u002Fgithub.com\u002Fprogsource\u002Fmaddy)\n- [WhisperSpeech](https:\u002F\u002Fcollabora.github.io\u002FWhisperSpeech\u002F)\n- [libxdo](https:\u002F\u002Fgithub.com\u002Fjordansissel\u002Fxdotool)\n- [Kokoro](https:\u002F\u002Fgithub.com\u002Fhexgrad\u002Fkokoro)\n- [Parler-TTS](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fparler-tts)\n- [F5-TTS](https:\u002F\u002Fgithub.com\u002FSWivid\u002FF5-TTS)\n- [S.A.M.](https:\u002F\u002Fdiscordier.github.io\u002Fsam)\n\n## Reviews and demos\n\n- [Speech Note 4.8 changes video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ww6skKOOzZ8) (Speech Note 4.8)\n- [Speech Note 4.7 changes video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=cEht4Fts6Bo) (Speech Note 4.7)\n- [Speech Note 4.6 changes video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=AVW5OY63wjg) (Speech Note 4.6)\n- [Speech Note 4.5 changes video](https:\u002F\u002Fyoutu.be\u002FS9MJ7y8-bcw) (Speech Note 4.5)\n- [Screenshots](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote\u002F-\u002Ftree\u002Fmain\u002Fdesktop\u002Fscreenshots) (Speech Note 4.5)\n\n- [Thejesh GN blog](https:\u002F\u002Fthejeshgn.com\u002F2025\u002F05\u002F26\u002Fspeech-note-app-for-offline-speech-to-text-tts-and-translation\u002F) (Speech Note 4.7)\n- [LinuxD0 video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=OCnCY49CmDQ) (Speech Note 4.7, Spanish)\n- [Guia Linux video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=_rL3II3JHNc) (Speech Note 4.7, Portuguese)\n- [lwn.net](https:\u002F\u002Flwn.net\u002FArticles\u002F987315\u002F) (Speech Note 4.6)\n- [Softpedia](https:\u002F\u002Flinux.softpedia.com\u002Fget\u002FUtilities\u002FSpeech-Note-104828.shtml) (Speech Note 4.6)\n- [OSTechNix](https:\u002F\u002Fostechnix.com\u002Fspeech-note-speech-recognition-text-to-speech-translation-app-for-linux\u002F) (Speech Note 4.6)\n- [Best FREE Speech-to-Text For Linux Mint video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=VDMbWUfHsbk) (Speech Note 4.6)\n- [Marco's Box](https:\u002F\u002Fwww.marcosbox.org\u002F2024\u002F02\u002Fspeech-note-trascrivi-e-traduci-offline-.html) (Speech Note 4.4, Italian)\n- [Marco's Box video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=6fNgZlh-O-w) (Speech Note 4.4, Italian)\n- [alternativalinux](https:\u002F\u002Fwww.alternativalinux.it\u002Friconoscimento-sintesi-vocale-e-traduttore-per-linux\u002F) (Speech Note 4.4, Italian)\n- [alternativalinux video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=6Peoss66fMg) (Speech Note 4.4, Italian)\n- [ZDNET](https:\u002F\u002Fwww.zdnet.com\u002Farticle\u002Fhow-to-enable-speech-to-text-in-linux-with-this-simple-app\u002F) (Speech Note 4.2)\n- [Translator feature video demo on Sailfish OS](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=88cdPpvBmmI) (Speech Note 4.0)\n- [Translator feature video demo on PinePhone](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=kTsM3kUxE2Q) (Speech Note 4.0)\n- [DebugPoint.com](https:\u002F\u002Fwww.debugpoint.com\u002Fspeech-note-text-to-speech\u002F) (Speech Note 4.0)\n- [DebugPoint.com video](https:\u002F\u002Fyoutu.be\u002FdYIPyS3F_eU) (Speech Note 4.0)\n- [OMG! Linux](https:\u002F\u002Fwww.omglinux.com\u002Fspeech-note-transcribe-voice-to-text-on-linux\u002F) (Speech Note 4.0)\n- [LinuxLinks](https:\u002F\u002Fwww.linuxlinks.com\u002Fmachine-learning-linux-speech-note\u002F) (Speech Note 4.0)\n- [The Linux Cast video](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=zlLVgTB42Bo) (Speech Note 4.0)\n- [CONNECTwww.com](https:\u002F\u002Fconnectwww.com\u002Fspeech-note-offline-speech-to-text-text-to-speech-and-translation-app\u002F) (Speech Note 4.0)\n\n## License\n\n**Speech Note** is an open source project. Source code is released under the\n[Mozilla Public License Version 2.0](https:\u002F\u002Fwww.mozilla.org\u002FMPL\u002F2.0\u002F). See the copy of the [LICENSE](.\u002FLICENSE)\n\n3rd party libraries:\n\n- **Coqui STT**, released under the\n[Mozilla Public License Version 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fcoqui-ai\u002FSTT\u002Fmain\u002FLICENSE)\n- **Coqui TTS**, released under the\n[Mozilla Public License Version 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fcoqui-ai\u002FTTS\u002Fdev\u002FLICENSE.txt)\n- **Vosk API**, released uder the [Apache License 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Falphacep\u002Fvosk-api\u002Fmaster\u002FCOPYING)\n- **whisper.cpp**, released under the [MIT License](https:\u002F\u002Fraw.githubusercontent.com\u002Fggerganov\u002Fwhisper.cpp\u002Fmaster\u002FLICENSE)\n- **WebRTC**, released under [this license](https:\u002F\u002Fwebrtc.googlesource.com\u002Fsrc\u002F+\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE)\n- **libarchive**, released under the [BSD License](https:\u002F\u002Fraw.githubusercontent.com\u002Flibarchive\u002Flibarchive\u002Fmaster\u002FCOPYING)\n- **RNNoise-nu**, released under the [BSD 3-Clause License](https:\u002F\u002Fraw.githubusercontent.com\u002FGregorR\u002Frnnoise-nu\u002Fmaster\u002FCOPYING)\n- **{fmt}**, released uder [this license](https:\u002F\u002Fraw.githubusercontent.com\u002Ffmtlib\u002Ffmt\u002Fmaster\u002FLICENSE.rst)\n- **Hugging Face Transformers**, released under the [Apache License 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fhuggingface\u002Ftransformers\u002Fmain\u002FLICENSE)\n- **Piper**, released under the [MIT License](https:\u002F\u002Fraw.githubusercontent.com\u002Frhasspy\u002Fpiper\u002Fmaster\u002FLICENSE.md)\n- **RHVoice**, released under the [GNU General Public License v2.0](https:\u002F\u002Fraw.githubusercontent.com\u002FRHVoice\u002FRHVoice\u002Fmaster\u002FLICENSE.md)\n- **ssplit-cpp**, released under the [Apache License 2.0](https:\u002F\u002Fgithub.com\u002Fugermann\u002Fssplit-cpp\u002Fraw\u002Fmaster\u002FLICENSE.md)\n- **espeak-ng**, released under the [GNU General Public License v3.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fespeak-ng\u002Fespeak-ng\u002Fmaster\u002FCOPYING)\n- **bergamot-translator**, released under the [Mozilla Public License 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fbrowsermt\u002Fbergamot-translator\u002Fmain\u002FLICENSE)\n- **Rubber Band Library**, released under the [GNU General Public License (version 2 or later)](https:\u002F\u002Fbreakfastquay.com\u002Frubberband\u002Flicense.html)\n- **simdjson**, released under the [Apache License 2.0](https:\u002F\u002Fgithub.com\u002Fsimdjson\u002Fsimdjson\u002Fraw\u002Fmaster\u002FLICENSE)\n- **Nlohmann JSON**, released under the [MIT License](https:\u002F\u002Fjson.nlohmann.me\u002Fhome\u002Flicense\u002F)\n- **uroman**, released under [this license](https:\u002F\u002Fgithub.com\u002Fisi-nlp\u002Furoman\u002Fraw\u002Fmaster\u002FLICENSE.txt)\n- **astrunc**, released under the [MIT License](https:\u002F\u002Fraw.githubusercontent.com\u002FJoke-Shi\u002Fastrunc\u002Fmaster\u002FLICENSE)\n- **FFmpeg**, released under the [GNU Lesser General Public License version 2.1 or later](https:\u002F\u002Fgit.ffmpeg.org\u002Fgitweb\u002Fffmpeg.git\u002Fblob_plain\u002FHEAD:\u002FLICENSE.md)\n- **LAME**, released under the LGPL\n- **Vorbis**, released under [this license](https:\u002F\u002Fgitlab.xiph.org\u002Fxiph\u002Fvorbis\u002F-\u002Fraw\u002Fmaster\u002FCOPYING?ref_type=heads)\n- **TagLib**, released under the [GNU Lesser General Public License (LGPL)](https:\u002F\u002Fraw.githubusercontent.com\u002Ftaglib\u002Ftaglib\u002Fmaster\u002FCOPYING.LGPL)\n and [Mozilla Public License (MPL)](https:\u002F\u002Fraw.githubusercontent.com\u002Ftaglib\u002Ftaglib\u002Fmaster\u002FCOPYING.MPL)\n- **libnumbertext**, released under the [BSD License](https:\u002F\u002Fraw.githubusercontent.com\u002FNumbertext\u002Flibnumbertext\u002Fmaster\u002FCOPYING)\n- **KDBusAddons**, released under the [LGPL licenses](https:\u002F\u002Finvent.kde.org\u002Fframeworks\u002Fkdbusaddons\u002F-\u002Ftree\u002Fmaster\u002FLICENSES?ref_type=heads)\n- **QHotkey**, released under the [BSD-3-Clause License](https:\u002F\u002Fraw.githubusercontent.com\u002FSkycoder42\u002FQHotkey\u002Fmaster\u002FLICENSE)\n- **faster-whisper**, released under the [MIT License](https:\u002F\u002Fgithub.com\u002Fguillaumekln\u002Ffaster-whisper\u002Fraw\u002Fmaster\u002FLICENSE)\n- **Mimic 3**, released under the [AGPL-3.0 license](https:\u002F\u002Fgithub.com\u002FMycroftAI\u002Fmimic3\u002Fraw\u002Fmaster\u002FLICENSE)\n- **Unikud**, released under the [MIT License](https:\u002F\u002Fgithub.com\u002Fmorrisalp\u002Funikud\u002Fraw\u002Fmain\u002FLICENSE)\n- **april-asr**, released under the [GNU General Public License v3.0](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr\u002Fraw\u002Fmain\u002FCOPYING)\n- **libopus**, released under [this license](https:\u002F\u002Fgitlab.xiph.org\u002Fxiph\u002Fopus\u002F-\u002Fraw\u002Fmaster\u002FCOPYING)\n- **html2md**, released under the [MIT License](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n- **maddy**, released under the [MIT License](https:\u002F\u002Fraw.githubusercontent.com\u002Fprogsource\u002Fmaddy\u002Fmaster\u002FLICENSE)\n- **WhisperSpeech**, released under the [MIT License](https:\u002F\u002Fraw.githubusercontent.com\u002Fcollabora\u002FWhisperSpeech\u002Fmain\u002FLICENSE)\n- **Kokoro**, released under the [Apache License 2.0](https:\u002F\u002Fgithub.com\u002Fhexgrad\u002Fkokoro\u002Fraw\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE)\n- **Parler-TTS**, released under the [Apache License 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fhuggingface\u002Fparler-tts\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE)\n- **F5-TTS**, released under the [MIT License](https:\u002F\u002Fgithub.com\u002FSWivid\u002FF5-TTS\u002Fraw\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE)\n\nThe files in the directory `nonbreaking_prefixes` were copied from\n[mosesdecoder](https:\u002F\u002Fgithub.com\u002Fmoses-smt\u002Fmosesdecoder) project and distributed under the\n[GNU Lesser General Public License v2.1](https:\u002F\u002Fgithub.com\u002Fmoses-smt\u002Fmosesdecoder\u002Fraw\u002Fmaster\u002FCOPYING).\n","# 语音笔记\n\n适用于 Linux 桌面和 Sailfish OS 的笔记、阅读和翻译应用，支持离线的语音转文本、文本转语音以及机器翻译功能。\n\n\u003Ca href='https:\u002F\u002Fflathub.org\u002Fapps\u002Fnet.mkiol.SpeechNote'>\u003Cimg width='240' alt='在 Flathub 上下载' src='https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmkiol_dsnote_readme_410555b9dfc2.png'\u002F>\u003C\u002Fa>\n\n## 本 README 的内容\n\n- [描述](#description)\n- [语言与模型](#languages-and-models)\n- [安装方法](#how-to-install)\n- [Flatpak 软件包](#flatpak-packages)\n- [测试版](#beta-version)\n- [额外功能](#extra-features)\n- [从源代码构建](#building-from-sources)\n- [如何启用自定义模型](#how-to-enable-a-custom-model)\n- [为 Speech Note 做贡献](#contributing-to-speech-note)\n- [如何支持](#how-to-support)\n- [评测与演示](#reviews-and-demos)\n- [许可证](#license)\n\n## 描述\n\n**Speech Note** 让您能够以多种语言记录、朗读和翻译笔记。它通过语音转文本、文本转语音和机器翻译来实现这些功能。文本和语音处理完全在本地进行，无需网络连接，您的隐私始终受到保护，没有任何数据会被发送到互联网。\n\n**Speech Note** 使用多种不同的处理引擎来完成其任务。目前使用的引擎包括：\n\n- 语音转文本 (STT)\n - [Coqui STT（Mozilla DeepSpeech 的分支）](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FSTT)\n - [Vosk](https:\u002F\u002Falphacephei.com\u002Fvosk)\n - [whisper.cpp](https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fwhisper.cpp)\n - [Faster Whisper](https:\u002F\u002Fgithub.com\u002Fguillaumekln\u002Ffaster-whisper)\n - [april-asr](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr)\n- 文本转语音 (TTS)\n - [espeak-ng](https:\u002F\u002Fgithub.com\u002Fespeak-ng\u002Fespeak-ng)\n - [MBROLA](https:\u002F\u002Fgithub.com\u002Fnumediart\u002FMBROLA)\n - [Piper](https:\u002F\u002Fgithub.com\u002Frhasspy\u002Fpiper)\n - [RHVoice](https:\u002F\u002Fgithub.com\u002FRHVoice\u002FRHVoice)\n - [Coqui TTS](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FTTS)\n - [Mimic 3](https:\u002F\u002Fmycroft.ai\u002Fmimic-3)\n - [WhisperSpeech](https:\u002F\u002Fcollabora.github.io\u002FWhisperSpeech\u002F)\n - [Kokoro](https:\u002F\u002Fgithub.com\u002Fhexgrad\u002Fkokoro)\n - [Parler-TTS](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fparler-tts)\n - [F5-TTS](https:\u002F\u002Fgithub.com\u002FSWivid\u002FF5-TTS)\n - [S.A.M.](https:\u002F\u002Fdiscordier.github.io\u002Fsam)\n- 机器翻译 (MT)\n - [Bergamot Translator](https:\u002F\u002Fgithub.com\u002Fbrowsermt\u002Fbergamot-translator)\n\n## 语言与模型\n\nSpeech Note 的安装包并不包含所支持模型的检查点文件，但用户可以通过应用程序内置的图形化模型浏览器轻松下载所需模型。\n\n以下语言及模型已被支持并可下载：\n\n| **语言ID** | **名称** | **DeepSpeech (语音识别)** | **Whisper (语音识别)** | **Vosk (语音识别)** | **April-ASR (语音识别)** | **Piper (文本转语音)** | **RHVoice (文本转语音)** | **espeak (文本转语音)** | **MBROLA (文本转语音)** | **Coqui (文本转语音)** | **Mimic3 (文本转语音)** | **WhisperSpeech (文本转语音)** | **Kokoro (文本转语音)** | **F5-TTS** | **Parler-TTS** | **S.A.M. (文本转语音)** | **Bergamot (机器翻译)** |\n| ----------- | ------------- | -------------------- | ----------------- | -------------- | ------------------- | --------------- | ----------------- | ---------------- | ---------------- | --------------- | ---------------- | ----------------------- | ---------------- | ---------- | -------------- | ---------------- | ----------------- |\n| af | 南非语 | | ● | | | | | ● | | | ● | | | | | | |\n| am | 阿姆哈拉语 | ● (e) | ● | | | | | ● | | ● | | | | | | | |\n| ar | 阿拉伯语 | | ● | ● | | ● | | ● | ● | ● | | | | | | | ● |\n| az | 阿塞拜疆语 | | ● | | | | | | | | | | | | | | ● |\n| be | 白俄罗斯语 | | ● | | | | | | | | | | | | | | ● |\n| bg | 保加利亚语 | | ● | | | | | ● | | ● | | | | | | | |\n| bn | 孟加拉语 | | ● | | | | | ● | | ● | ● | | | | | | |\n| bs | 波斯尼亚语 | | ● | | | | | ● | | | | | | | | | ● |\n| ca | 加泰罗尼亚语 | ● | ● | ● | | ● | | ● | | ● | | | | | | | ● |\n| cs | 捷克语 | ● | ● | ● | | ● | ● | ● | ● | ● | | | | | | | ● |\n| cy | 威尔士语 | | | | | ● | | | | | | | | | | | |\n| da | 丹麦语 | | ● | | | ● | | ● | | ● | | | | | | | ● |\n| de | 德语 | ● | ● | ● | | ● | | ● | | ● | ● | ● | | | ●(e) | | ● |\n| el | 希腊语 | ● (e) | ● | | | ● | | ● | | ● | ● | | | | | | ● |\n| en | 英语 | ● | ● | ● | ● | ● | ● | ● | | ● | ● | ● | ● | ● | ● | ● | ● |\n| eo | 世界语 | | | ● | | | ● | ● | | | | | | | | | |\n| es | 西班牙语 | ● | ● | ● | | ● | ● | ● | | ● | ● | ● | ● | | ●(e) | | ● |\n| et | 爱沙尼亚语 | ● (e) | ● | | | | | ● | ● | ● | | | | | | | ● |\n| eu | 巴斯克语 | ● (e) | ● | | | | | ● | | ● | | | | | | | |\n| fa | 波斯语 | ● | ● | ● | | ● | | ● | ● | ● | ● | | | | | | ● |\n| fi | 芬兰语 | ● | ● | | | ● | | ● | | ● | ● | | | | | | ● |\n| fr | 法语 | ● | ● | ● | ● | ● | | ● | | ● | ● | ● | ● | | ●(e) | | ● |\n| ga | 爱尔兰语 | | | | | | | ● | | ● | | | | | | | |\n| gu | 古吉拉特语 | | ● | | | | | ● | | | ● | | | | | | |\n| ha | 豪萨语 | | ● | | | | | | | | ● | | | | | | |\n| he | 希伯来语 | | ● | | | | | | | ● | | | | | | | |\n| hi | 印地语 | | ● | ● | | ● | | ● | | | | | ● | | | | |\n| hr | 克罗地亚语 | | ● | | | | ● | ● | ● | ● | | | | | | | |\n| hu | 匈牙利语 | ● (e) | ● | | | ● | | ● | ● | ● | ● | | | | | | ● |\n| id | 印度尼西亚语 | ● (e) | ● | | | | | ● | ● | ● | | | | | | | ● |\n| is | 冰岛语 | | ● | | | ● | | ● | | ● | | | | | | | ● |\n| it | 意大利语 | ● | ● | ● | | ● | | ● | | ● | ● | ● | ● | | ●(e) | | ● |\n| ja | 日语 | | ● | ● | | | | ● | | ● | | | ● | | | | ● |\n| jv | 爪哇语 | | ● | | | | | | | | ● | | | | | | |\n| ka | 格鲁吉亚语 | | ● | | | ● | ● | ● | | | | | | | | | |\n| kk | 哈萨克语 | | ● | ● | | ● | | ● | | ● | | | | | | | |\n| kn | 卡纳达语 | | ● | | | | | | | ● | | | | | | | ● |\n| ko | 韩语 | | ● | ● | | | | ● | | ● | | | | | | | ● |\n| ky | 吉尔吉斯语 | | | | | | ● | ● | | | | | | | | | |\n| la | 拉丁语 | | | | | | | ● | | ● | | | | | | | |\n| lb | 卢森堡语 | | | | | ● | | | | | | | | | | | |\n| lt | 立陶宛语 | | ● | | | | | ● | ● | ● | | | | | | | ● |\n| lv | 拉脱维亚语 | ● | ● | | | ● | | ● | | ● | | | | | | | ● |\n| mk | 马其顿语 | | ● | | | | ● | ● | | | | | | | | | |\n| ml | 马拉雅拉姆语 | | ● (e) | | | ● | | | | ● | | | | | | | ● |\n| mn | 蒙古语 | ● (e) | ● | | | | | | | ● | | | | | | | |\n| mr | 马拉地语 | | ● | | | | | | | ● | | | | | | | |\n| ms | 马来语 | | ● | | | | | ● | ● | ● | | | | | | | |\n| mt | 马耳他语 | | ● | | | | | ● | | ● | | | | | | | |\n| ne | 尼泊尔语 | | ● | | | ● | | ● | | | ● | | | | | | |\n| nl | 荷兰语 | ● (e) | ● | ● | | ● | | ● | | ● | ● | ● | | | ●(e) | | ● |\n| no | 挪威语 | | ● | | | ● | | ● | | | | | | | | | ● |\n| pl | 波兰语 | ● | ● | ● | ● | ● | ● | ● | ● | ● | ● | ● | | | ●(e) | | ● |\n| pt | 葡萄牙语 | ● (e) | ● | ● | | ● | | ● | ● | ● | | | ● | | ●(e) | | ● |\n| ro | 罗马尼亚语 | ● (e) | ● | | | ● | | ● | ● | ● | | | | | | | ● |\n| ru | 俄语 | ● | ● | ● | | ● | ● | ● | | | ● | | | | | | ● |\n| sk | 斯洛伐克语 | | ● | | | ● | ● | ● | | ● | | | | | | | ● |\n| sl | 斯洛文尼亚语 | ● (e) | ● | | | ● | | ● | | ● | | | | | | | ● |\n| sq | 阿尔巴尼亚语 | | ● | | | | ● | ● | | ● | | | | | | | |\n| sr | 塞尔维亚语 | | ● | | | ● | ● | ● | | | | | | | | | ● |\n| sv | 瑞典语 | | ● | ● | | ● | | ● | ● | ● | ● | | | | | | ● |\n| sw | 斯瓦希里语 | ● | ● | | | ● | | ● | | ● | | | | | | | |\n| ta | 泰米尔语 | | ● | | | | | | | ● | | | | | | | ● |\n| te | 泰卢固语 | | ● | | | | | ● | | | ● | | | | | | |\n| th | 泰语 | ● (e) | ● | | | | | ● | | ● | | | | | | | |\n| tl | 他加禄语 | | ● | ● | | | | | | ● | | | | | | | |\n| tn | 茨瓦纳语 | | ● | | | | | ● | | | ● | | | | | | |\n| tr | 土耳其语 | ● (e) | ● | ● | | ● | | ● | ● | ● | | | | | | | ● |\n| tt | 鞑靼语 | | ● | | | | ● | ● | | ● | | | | | | | |\n| uk | 乌克兰语 | ● | ● | ● | | ● | ● | ● | | ● | ● | | | | | | ● |\n| uz | 乌兹别克语 | | ● | ● | | | | ● | | ● | | | | | | | |\n| vi | 越南语 | | ● | ● | | ● | | ● | | ● | | | | | | | ● |\n| yo | 约鲁巴语 | ● (e) | ● | | | | | | | ● | ● | | | | | | |\n| zh | 中文 | ● | ● | ● | | ● | | ● | | ● | | | ● | ● | | | ● |\n\n\u003Csup>(e) 实验性，很可能效果不佳\u003C\u002Fsup>\n\u003Cbr\u002F>\n\n更快的 Whisper、Coqui TTS 和 Mimic3 模型仅适用于 x86-64 架构。\n\n语言模型可以直接从应用程序中下载。\n\n当前已配置可供下载的模型详情，请参阅\n[models.json (GitHub)](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Fblob\u002Fmain\u002Fconfig\u002Fmodels.json) 或\n[models.json (GitLab)](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote\u002F-\u002Fblob\u002Fmain\u002Fconfig\u002Fmodels.json)。\n\n\n\n## 安装方法\n\n- Linux 桌面版：[Flatpak](https:\u002F\u002Fflathub.org\u002Fapps\u002Fnet.mkiol.SpeechNote)\n\n```sh\n# Flatpak 基础包\nflatpak install net.mkiol.SpeechNote\n\n# 可选的 NVIDIA 插件包\nflatpak install net.mkiol.SpeechNote.Addon.nvidia\n\n# 可选的 AMD 插件包\nflatpak install net.mkiol.SpeechNote.Addon.amd\n```\n\n- Arch Linux (AUR)：\n\n - [dsnote](https:\u002F\u002Faur.archlinux.org\u002Fpackages\u002Fdsnote)\n - [dsnote-git](https:\u002F\u002Faur.archlinux.org\u002Fpackages\u002Fdsnote-git)\n\n- openSUSE ([Packman 仓库](https:\u002F\u002Fen.opensuse.org\u002FAdditional_package_repositories#Packman))\n\n```sh\n# 基础包\nzypper in speechnote\n\n# 可选的对 Speech Note 中基于 Python 功能的支持\nzypper in speechnote-python-modules\n```\n\n- Sailfish OS：[OpenRepos](https:\u002F\u002Fopenrepos.net\u002Fcontent\u002Fmkiol\u002Fspeech-note)\n\n### Flatpak 软件包\n\n通过 Flatpak 分发的应用程序（发布在 Flathub 上）由以下软件包组成：\n\n- 基础包“Speech Note”（net.mkiol.SpeechNote）\n- 针对 NVIDIA 显卡的可选插件“Speech Note NVIDIA”（net.mkiol.SpeechNote.Addon.nvidia）\n- 针对 AMD 显卡的可选插件“Speech Note AMD”（net.mkiol.SpeechNote.Addon.amd）\n\n基础包包含了运行应用程序所有功能所需的所有依赖项。\n插件则增加了 GPU 加速功能，从而加快应用程序中某些操作的速度。\n\n基础包和插件都包含许多“重量级”库，如 CUDA、ROCm、Torch 和 Python 库等。\n因此，这些软件包的大小以及安装后所需的磁盘空间都非常大。\n如果您不需要全部功能，可以使用体积小得多的“Tiny”版本（可在 [Releases](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Freleases) 页面获取），它仅提供基本功能。如果需要，您也可以将“Tiny”版本与 GPU 加速插件一起使用。\n\n基础包、Tiny 版本和插件的 Flatpak 软件包大小对比：\n\n| **大小** | **基础包** | **Tiny 版本** | **AMD 插件** | **NVIDIA 插件** |\n| ------------- | ---------| --------------| --------------| -----------------|\n| 下载大小 | 1.2 GiB | 48 MiB | +7.6 GiB | +4.3 GiB |\n| 解压后大小 | 3.6 GiB | 170 MiB | +34 GiB | +9 GiB |\n\n插件使用所谓的“额外数据”。这意味着在安装过程中，会从原始来源下载几个二进制模块，并将其合并到 Flatpak 包中。由于这一过程，安装插件时会暂时需要额外的磁盘空间。请确保您安装 Flatpak 应用程序的分区有足够的可用空间。当 Flatpak 包以系统范围方式安装时（默认设置），应用程序会被存储在 _\u002Fvar\u002Flib\u002Fflatpak_ 中。如果是用户级别的安装（--user），应用程序则会被存储在 _~\u002F.local\u002Fshare\u002Fflatpak_ 中。\n\n安装插件所需的磁盘空间：\n\n| **插件** | **大小** |\n| --------------| ---------|\n| AMD 插件 | 55 GiB |\n| NVIDIA 插件 | 15 GiB |\n\n基础包、Tiny 版本和插件的特性对比：\n\n| **功能** | **基础包** | **Tiny 版本** | **AMD 插件** | **NVIDIA 插件** |\n| ----------------------------------- | ---------| --------------| --------------| -----------------|\n| Coqui\u002FDeepSpeech STT | + | + | | |\n| Vosk STT | + | + | | |\n| Whisper (whisper.cpp) STT | + | + | | |\n| Whisper (whisper.cpp) STT OpenCL ROCm | - | - | + | |\n| Whisper (whisper.cpp) STT OpenCL NVIDIA | + | + | | |\n| Whisper (whisper.cpp) STT ROCm | - | - | + | |\n| Whisper (whisper.cpp) STT CUDA | - | - | | + |\n| Whisper (whisper.cpp) STT OpenVINO | + | - | | |\n| Whisper (whisper.cpp) STT Vulkan | + | + | | |\n| FasterWhisper STT | + | - | | |\n| FasterWhisper STT CUDA | - | - | | + |\n| April-ASR STT | + | + | | |\n| eSpeak TTS | + | + | | |\n| MBROLA TTS | + | + | | |\n| Piper TTS | + | + | | |\n| RHVoice TTS | + | + | | |\n| Coqui TTS | + | - | | |\n| Coqui TTS ROCm | - | - | + | |\n| Coqui TTS CUDA | - | - | | + |\n| Mimic3 TTS | + | - | | |\n| WhisperSpeech TTS | + | - | | |\n| WhisperSpeech TTS ROCm | - | - | + | |\n| WhisperSpeech TTS CUDA | - | - | | + |\n| Kokoro TTS | + | - | | |\n| Kokoro TTS ROCm | - | - | + | |\n| Kokoro TTS CUDA | - | - | | + |\n| Parler-TTS | + | - | | |\n| Parler-TTS ROCm | - | - | + | |\n| Parler-TTS CUDA | - | - | | + |\n| F5-TTS | + | - | | |\n| F5-TTS ROCm | - | - | + | |\n| F5-TTS CUDA | - | - | | + |\n| S.A.M TTS | + | + | | |\n| 标点符号恢复 | + | - | | |\n| 翻译 | + | + | | |\n\n### 测试版\n\n除了 Flathub 仓库中的稳定版本外，您还可以尝试测试即将发布的“Beta”版本。该版本虽然可用，但可能包含更多 bug。\n\nBeta 版本可在“flathub-beta”仓库中找到。请按照[这些说明](https:\u002F\u002Fdiscourse.flathub.org\u002Ft\u002Fhow-to-use-flathub-beta\u002F2111)在您的计算机上启用 flathub-beta。\n\n## 额外功能\n\n### 命令行选项\n\nCLI 界面主要用于在 _Speech Note_ 已经运行时与桌面集成（例如，隐藏在系统托盘或后台）。\n\n#### 示例\n\n列出所有支持的选项：\n\n```shell\nflatpak run net.mkiol.SpeechNote --help\n```\n\n开始监听：\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-listening\n```\n\n取消任何已启动的操作：\n\n```sh\nflatpak run net.mkiol.SpeechNote --action cancel\n```\n\n开始监听，解码后的文本将保存到剪贴板：\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-listening-clipboard\n```\n\n开始监听，解码后的文本将插入到桌面上光标所在的任何窗口中：\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-listening-active-window\n```\n\n开始朗读“Hello, how are you doing?”：\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-reading-text --text \"Hello, how are you doing?\"\n```\n\n将“Hello, how are you doing?”的语音保存到“speech.mp3”文件中：\n\n```sh\nflatpak run net.mkiol.SpeechNote --action start-reading-text --text \"Hello, how are you doing?\" --output-file speech.mp3\n```\n\n列出所有可用的 TTS 模型：\n\n```sh\nflatpak run net.mkiol.SpeechNote --print-available-models tts\n```\n\n### 全局键盘快捷键\n\n全局键盘快捷键允许您使用键盘启动监听或朗读功能，即使应用程序未处于活动状态（例如最小化、隐藏在系统托盘图标中或仅在后台运行）。\n\n要启用并自定义键盘快捷键，请前往 `设置`->`辅助功能`->`使用全局键盘快捷键`。\n\n为了使快捷键在 Wayland 下正常工作，您的桌面环境必须在 _XDG Desktop Portal_ 服务中支持 [GlobalShortcuts 接口](https:\u002F\u002Fflatpak.github.io\u002Fxdg-desktop-portal\u002Fdocs\u002Fdoc-org.freedesktop.impl.portal.GlobalShortcuts.html)。目前，`GlobalShortcuts` 仅在最新的 KDE Plasma 和 GNOME 桌面环境中受支持。\n\n当使用 `XDG Desktop Portal` 来管理全局快捷键时，请使用桌面环境自带的工具来自定义按键绑定。\n\n### 插入到当前窗口\n\n通过全局键盘快捷键或命令行操作，您可以直接开始监听并将解码后的文本插入到当前焦点窗口中。这使得您能够将 _Speech Note_ 用作桌面端的语音输入工具。\n\n在 X11 下，此功能应可开箱即用。\n\n在 Wayland 下，必须安装并运行外部 `ydotool` 守护进程才能使其正常工作。如果您使用 Flatpak，请确保应用程序具有访问 `ydotool` 守护进程套接字文件及其所在目录的权限。\n\n### 字幕\n\n在进行语音转文字 (STT) 时，您也可以将音频转换为字幕而不是纯文本。支持的格式是 SRT。\n\nTTS 功能也支持字幕输出。朗读字幕时会根据字幕的时间戳进行同步。如果进行了相应配置，语音速度可以自动调整以匹配字幕片段的持续时间。这样就可以从字幕文件生成旁白。\n\n## 从源代码构建\n\n### Arch Linux\n\n您还可以使用提供的 PKGBUILD 文件从仓库构建并安装最新的开发版（git）或最新稳定版（release）（请注意，以下关于在 Linux 上构建的说明同样适用）：\n\n```sh\ngit clone \u003Cgit 仓库 URL>\n\ncd dsnote\u002Farch\u002Fgit # 构建最新 git 版本\n# 或\ncd dsnote\u002Farch\u002Frelease # 构建最新 release 版本\n\nmakepkg -si\n```\n\n### RHEL\u002FFedora\u002FRocky Linux\n\n您也可以使用提供的 SPEC 文件和辅助脚本 `make_rpm.sh` 从仓库构建并安装最新的开发版本：\n\n```sh\ngit clone \u003Cgit 仓库 URL>\n\ncd dsnote\u002Ffedora\n\n# 可选：安装构建依赖\ndnf install rpmdevtools autoconf automake boost-devel cmake git kf5-kdbusaddons-devel libarchive-devel libxdo-devel libXinerama-devel libxkbcommon-x11-devel libXtst-devel libtool meson openblas-devel patchelf pybind11-devel python3-devel python3-pybind11 qt5-linguist qt5-qtmultimedia-devel qt5-qtquickcontrols2-devel qt5-qtx11extras-devel rubberband-devel taglib-devel vulkan-headers\n\n.\u002Fmake_rpm.sh\n```\n\n### Flatpak\n\n```sh\ngit clone \u003Cgit 仓库 URL>\n\ncd dsnote\u002Fflatpak\n\n# 构建基础包\nflatpak-builder --force-clean --user --install-deps-from=flathub --repo=\"\u003C名称或本地 flatpak 仓库路径>\" \"\u002F输出目录路径\" net.mkiol.SpeechNote.yaml\n\n# 构建可选的 NVIDIA 插件包\nflatpak-builder --force-clean --user --install-deps-from=flathub --repo=\"\u003C名称或本地 flatpak 仓库路径>\" \"\u002F输出目录路径\" net.mkiol.SpeechNote.Addon.nvidia.yaml\n```\n\n### Sailfish OS\n\n```sh\ngit clone \u003Cgit 仓库 URL>\n\ncd dsnote\nmkdir build\ncd build\n\nsfdk config --session specfile=..\u002Fsfos\u002Fharbour-dsnote.spec\nsfdk config --session target=SailfishOS-4.4.0.58-aarch64\nsfdk cmake ..\u002F -DCMAKE_BUILD_TYPE=Release -DWITH_SFOS=ON -DWITH_PY=OFF\nsfdk package\n```\n\n### Linux（直接构建）\n\nSpeech Note 有许多编译时和运行时依赖项，包括共享库和静态库、第三方可执行文件以及 Python 和 Perl 脚本。由于这些复杂性，推荐的构建方式是使用 Flatpak 工具链（Flatpak 清单文件和 [flatpak-builder](https:\u002F\u002Fdocs.flatpak.org\u002Fen\u002Flatest\u002Fflatpak-builder.html)）。如果您希望直接构建（即不使用 Flatpak），也是可行的，但会更加复杂。\n\n```sh\ngit clone \u003Cgit 仓库 URL>\n\ncd dsnote\nmkdir build\ncd build\n\ncmake ..\u002F -DCMAKE_BUILD_TYPE=Release -DWITH_DESKTOP=ON\nmake\n```\n\n若要构建时不包含 Python 组件，可在 CMake 配置步骤中添加 `-DWITH_PY=OFF`。\n\n如需了解其他构建选项，请在 `CMakeList.txt` 文件中搜索 `option(BUILD_XXX)`。\n\n### 最小化 Linux 构建\n\n为了在本地开发功能，完整构建过程可能会耗费大量时间；您可以使用以下方法进行最小化直接构建：\n\n```sh\ngit clone \u003Cgit 仓库 URL>\n\ncd dsnote\nmkdir build\ncd build\n\ncmake ..\u002F -DWITH_DESKTOP=ON \\\n -DWITH_PY=ON \\\n -DDOWNLOAD_VOSK=ON \\\n -DBUILD_VOSK=OFF \\\n -DBUILD_WHISPERCPP=OFF \\\n -DBUILD_OPENBLAS=OFF \\\n -DBUILD_RHVOICE=OFF \\\n -DBUILD_RHVOICE_MODULE=OFF \\\n -DBUILD_BERGAMOT=OFF\nmake\n```\n\n## 构建 WlClipboard\n\n如果您想构建 wl-clipboard，或者您的系统中没有 wl-clipboard，并且在尝试将文本粘贴到当前窗口时，KDE Klipper 或 QClipboard 无法正常工作，您可以添加以下标志来构建 wl-clipboard：\n\n```sh\ncmake ..\u002F -DWITH_DESKTOP=ON -DCMAKE_BUILD_TYPE=Release -DBUILD_WL_CLIPBOARD\n```\n\n## 如何启用自定义模型\n\n所有可供下载的模型均在配置文件（config\u002Fmodels.json）中指定。要启用与当前支持的引擎兼容的自定义模型，只需编辑此文件并重启应用程序即可。\n\n首次运行应用程序时，模型配置文件会创建在以下路径：\n\n- `~\u002F.local\u002Fshare\u002Fnet.mkiol\u002Fdsnote\u002Fmodels.json`，或\n- `~\u002F.var\u002Fapp\u002Fnet.mkiol.SpeechNote\u002Fdata\u002Fnet.mkiol\u002Fdsnote\u002Fmodels.json`（Flatpak），或\n- `~\u002F.local\u002Fshare\u002Forg.mkiol\u002Fdsnote\u002Fmodels.json`（Sailfish OS）\n\n您可以自由编辑当前已启用的模型，也可以添加新的模型。\n\n模型定义如下：\n\n```json\n{\n \"name\": \"\u003C模型名称>\",\n \"model_id\": \"\u003C模型唯一标识>\",\n \"engine\": \"\u003C引擎类型>\",\n \"lang_id\": \"\u003C语言标识>\",\n \"checksum\": \"\u003CMD5校验和>\",\n \"checksum_quick\": \"\u003C部分MD5校验和>\",\n \"comp\": \"\u003C压缩类型>\",\n \"urls\": [\n \u003C模型URLs>\n ],\n \"size\": \"\u003C所有文件的下载大小>\"\n}\n```\n\n允许的引擎类型：`stt_ds`、`stt_vosk`、`stt_april`、`stt_whisper`、`stt_fasterwhisper`、`tts_piper`、`tts_rhvoice`、`tts_espeak`、`tts_coqui`、`tts_mimic3`、`tts_whisperspeech`、`tts_sam`、`tts_parler`、`tts_f5`、`tts_kokoro`、`mnt_bergamot`\n\n允许的压缩类型：`none`、`gz`、`xz`、`tarxz`、`targz`、`zip`、`zipall`、`dir`、`dirgz`\n\n允许的URL类型：`http`、`https`、`file`\n\n校验和是在解压所有文件后计算的。如果您要添加新模型，可以使用命令行选项 `--gen-checksums` 来获取正确的校验和。为此，请将 `checksum` 和 `checksum_quick` 字段留空，保存文件，并使用上述选项运行 Speech Note。\n\n例如：\n\n```json\n{\n \"name\": \"新Piper语音\",\n \"model_id\": \"en_piper_new\",\n \"engine\": \"tts_piper\",\n \"lang_id\": \"en\",\n \"checksum\": \"\",\n \"checksum_quick\": \"\",\n \"size\": \"\"\n \"comp\": \"dir\",\n \"urls\": [\n \"file:\u002F\u002F\u002Fhome\u002Fme\u002Fmodels\u002Fnew-model-medium.onnx\",\n \"file:\u002F\u002F\u002Fhome\u002Fme\u002Fmodels\u002Fnew-model-medium.onnx.json\"\n ]\n}\n```\n\n```sh\nflatpak run net.mkiol.SpeechNote --verbose --gen-checksums\n```\n\n## 参与Speech Note开发\n\n我们非常欢迎任何形式的贡献！\n\n有关如何参与开发的详细信息，请参阅 [CONTRIBUTING.md](CONTRIBUTING.md)。\n\n该项目同时托管在 [GitHub](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote) 和 [GitLab](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote) 上。您可以在自己最常用的平台上提交PR\u002FMR、报告问题或请求新功能。\n\n### 翻译\n\nQt格式的翻译文件位于 `translations` 目录下。\n\n推荐通过 [Transifex服务](https:\u002F\u002Fexplore.transifex.com\u002Fmkiol\u002Fdsnote\u002F) 提交翻译，但如果您希望直接提交PR\u002FMR，也完全可以。\n\n## 如何支持\n\n如果您觉得 **Speech Note** 非常有用，并希望支持这个项目，请考虑采取以下一种或多种方式：\n\n- 在 [GitHub](https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote) 或\u002F及 [GitLab](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote) 上给个赞。\n- 在您的应用管理器（Discover、Software或其他）中写下评论。\n- 通过社交媒体向他人推荐这款应用。\n- 如果您有闲钱，可以通过 [ko-fi（一次性捐赠）](https:\u002F\u002Fko-fi.com\u002Fmkiol) 或 [Liberapay（定期捐赠）](https:\u002F\u002Fliberapay.com\u002Fmkiol\u002Fdonate) 进行小额捐助。\n\n## 依赖库\n\n**Speech Note** 依赖于以下开源项目：\n\n- [Qt](https:\u002F\u002Fwww.qt.io\u002F)\n- [Coqui STT](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FSTT)\n- [Coqui TTS](https:\u002F\u002Fgithub.com\u002Fcoqui-ai\u002FTTS)\n- [Vosk](https:\u002F\u002Falphacephei.com\u002Fvosk)\n- [whisper.cpp](https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fwhisper.cpp)\n- [WebRTC VAD](https:\u002F\u002Fwebrtc.org\u002F)\n- [libarchive](https:\u002F\u002Flibarchive.org\u002F)\n- [RNNoise-nu](https:\u002F\u002Fgithub.com\u002FGregorR\u002Frnnoise-nu)\n- [{fmt}](https:\u002F\u002Ffmt.dev)\n- [Hugging Face Transformers](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Ftransformers)\n- [Piper](https:\u002F\u002Fgithub.com\u002Frhasspy\u002Fpiper)\n- [RHVoice](https:\u002F\u002Fgithub.com\u002FRHVoice\u002FRHVoice)\n- [ssplit-cpp](https:\u002F\u002Fgithub.com\u002Fugermann\u002Fssplit-cpp)\n- [espeak-ng](https:\u002F\u002Fgithub.com\u002Fespeak-ng\u002Fespeak-ng)\n- [bergamot-translator](https:\u002F\u002Fgithub.com\u002Fbrowsermt\u002Fbergamot-translator)\n- [Rubber Band Library](https:\u002F\u002Fbreakfastquay.com\u002Frubberband)\n- [simdjson](https:\u002F\u002Fsimdjson.org\u002F)\n- [Nlohmann JSON](https:\u002F\u002Fjson.nlohmann.me\u002F)\n- [uroman](https:\u002F\u002Fgithub.com\u002Fisi-nlp\u002Furoman)\n- [astrunc](https:\u002F\u002Fgithub.com\u002FJoke-Shi\u002Fastrunc)\n- [FFmpeg](https:\u002F\u002Fffmpeg.org\u002F)\n- [LAME](https:\u002F\u002Flame.sourceforge.io\u002F)\n- [Vorbis](https:\u002F\u002Fxiph.org\u002Fvorbis\u002F)\n- [TagLib](https:\u002F\u002Ftaglib.org\u002F)\n- [libnumbertext](https:\u002F\u002Fgithub.com\u002FNumbertext\u002Flibnumbertext)\n- [KDBusAddons](https:\u002F\u002Finvent.kde.org\u002Fframeworks\u002Fkdbusaddons)\n- [QHotkey](https:\u002F\u002Fgithub.com\u002FSkycoder42\u002FQHotkey)\n- [faster-whisper](https:\u002F\u002Fgithub.com\u002Fguillaumekln\u002Ffaster-whisper)\n- [Mimic 3](https:\u002F\u002Fmycroft.ai\u002Fmimic-3)\n- [Unikud](https:\u002F\u002Fgithub.com\u002Fmorrisalp\u002Funikud)\n- [april-asr](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr)\n- [Opus](https:\u002F\u002Fopus-codec.org\u002F)\n- [html2md](https:\u002F\u002Ftim-gromeyer.github.io\u002Fhtml2md\u002F)\n- [maddy](https:\u002F\u002Fgithub.com\u002Fprogsource\u002Fmaddy)\n- [WhisperSpeech](https:\u002F\u002Fcollabora.github.io\u002FWhisperSpeech\u002F)\n- [libxdo](https:\u002F\u002Fgithub.com\u002Fjordansissel\u002Fxdotool)\n- [Kokoro](https:\u002F\u002Fgithub.com\u002Fhexgrad\u002Fkokoro)\n- [Parler-TTS](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fparler-tts)\n- [F5-TTS](https:\u002F\u002Fgithub.com\u002FSWivid\u002FF5-TTS)\n- [S.A.M.](https:\u002F\u002Fdiscordier.github.io\u002Fsam)\n\n## 评论和演示\n\n- [Speech Note 4.8 变更视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ww6skKOOzZ8) (Speech Note 4.8)\n- [Speech Note 4.7 变更视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=cEht4Fts6Bo) (Speech Note 4.7)\n- [Speech Note 4.6 变更视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=AVW5OY63wjg) (Speech Note 4.6)\n- [Speech Note 4.5 变更视频](https:\u002F\u002Fyoutu.be\u002FS9MJ7y8-bcw) (Speech Note 4.5)\n- [截图](https:\u002F\u002Fgitlab.com\u002Fmkiol\u002Fdsnote\u002F-\u002Ftree\u002Fmain\u002Fdesktop\u002Fscreenshots) (Speech Note 4.5)\n\n- [Thejesh GN 博客](https:\u002F\u002Fthejeshgn.com\u002F2025\u002F05\u002F26\u002Fspeech-note-app-for-offline-speech-to-text-tts-and-translation\u002F) (Speech Note 4.7)\n- [LinuxD0 视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=OCnCY49CmDQ) (Speech Note 4.7，西班牙语)\n- [Guia Linux 视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=_rL3II3JHNc) (Speech Note 4.7，葡萄牙语)\n- [lwn.net](https:\u002F\u002Flwn.net\u002FArticles\u002F987315\u002F) (Speech Note 4.6)\n- [Softpedia](https:\u002F\u002Flinux.softpedia.com\u002Fget\u002FUtilities\u002FSpeech-Note-104828.shtml) (Speech Note 4.6)\n- [OSTechNix](https:\u002F\u002Fostechnix.com\u002Fspeech-note-speech-recognition-text-to-speech-translation-app-for-linux\u002F) (Speech Note 4.6)\n- [Linux Mint 上最佳免费语音转文本工具视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=VDMbWUfHsbk) (Speech Note 4.6)\n- [Marco's Box](https:\u002F\u002Fwww.marcosbox.org\u002F2024\u002F02\u002Fspeech-note-trascrivi-e-traduci-offline-.html) (Speech Note 4.4，意大利语)\n- [Marco's Box 视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=6fNgZlh-O-w) (Speech Note 4.4，意大利语)\n- [alternativalinux](https:\u002F\u002Fwww.alternativalinux.it\u002Friconoscimento-sintesi-vocale-e-traduttore-per-linux\u002F) (Speech Note 4.4，意大利语)\n- [alternativalinux 视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=6Peoss66fMg) (Speech Note 4.4，意大利语)\n- [ZDNET](https:\u002F\u002Fwww.zdnet.com\u002Farticle\u002Fhow-to-enable-speech-to-text-in-linux-with-this-simple-app\u002F) (Speech Note 4.2)\n- [在 Sailfish OS 上的翻译功能视频演示](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=88cdPpvBmmI) (Speech Note 4.0)\n- [在 PinePhone 上的翻译功能视频演示](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=kTsM3kUxE2Q) (Speech Note 4.0)\n- [DebugPoint.com](https:\u002F\u002Fwww.debugpoint.com\u002Fspeech-note-text-to-speech\u002F) (Speech Note 4.0)\n- [DebugPoint.com 视频](https:\u002F\u002Fyoutu.be\u002FdYIPyS3F_eU) (Speech Note 4.0)\n- [OMG! Linux](https:\u002F\u002Fwww.omglinux.com\u002Fspeech-note-transcribe-voice-to-text-on-linux\u002F) (Speech Note 4.0)\n- [LinuxLinks](https:\u002F\u002Fwww.linuxlinks.com\u002Fmachine-learning-linux-speech-note\u002F) (Speech Note 4.0)\n- [The Linux Cast 视频](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=zlLVgTB42Bo) (Speech Note 4.0)\n- [CONNECTwww.com](https:\u002F\u002Fconnectwww.com\u002Fspeech-note-offline-speech-to-text-text-to-speech-and-translation-app\u002F) (Speech Note 4.0)\n\n## 许可证\n\n**Speech Note** 是一个开源项目。源代码根据\n[Mozilla 公共许可证 2.0 版](https:\u002F\u002Fwww.mozilla.org\u002FMPL\u002F2.0\u002F) 发布。请参阅 [LICENSE](.\u002FLICENSE) 文件的副本。\n\n第三方库：\n\n- **Coqui STT**，根据\n[Mozilla 公共许可证 2.0 版](https:\u002F\u002Fraw.githubusercontent.com\u002Fcoqui-ai\u002FSTT\u002Fmain\u002FLICENSE) 发布\n- **Coqui TTS**，根据\n[Mozilla 公共许可证 2.0 版](https:\u002F\u002Fraw.githubusercontent.com\u002Fcoqui-ai\u002FTTS\u002Fdev\u002FLICENSE.txt) 发布\n- **Vosk API**，根据 [Apache 许可证 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Falphacep\u002Fvosk-api\u002Fmaster\u002FCOPYING) 发布\n- **whisper.cpp**，根据 [MIT 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002Fggerganov\u002Fwhisper.cpp\u002Fmaster\u002FLICENSE) 发布\n- **WebRTC**，根据 [此许可证](https:\u002F\u002Fwebrtc.googlesource.com\u002Fsrc\u002F+\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE) 发布\n- **libarchive**，根据 [BSD 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002Flibarchive\u002Flibarchive\u002Fmaster\u002FCOPYING) 发布\n- **RNNoise-nu**，根据 [BSD 3-Clause 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002FGregorR\u002Frnnoise-nu\u002Fmaster\u002FCOPYING) 发布\n- **{fmt}**，根据 [此许可证](https:\u002F\u002Fraw.githubusercontent.com\u002Ffmtlib\u002Ffmt\u002Fmaster\u002FLICENSE.rst) 发布\n- **Hugging Face Transformers**，根据 [Apache 许可证 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fhuggingface\u002Ftransformers\u002Fmain\u002FLICENSE) 发布\n- **Piper**，根据 [MIT 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002Frhasspy\u002Fpiper\u002Fmaster\u002FLICENSE.md) 发布\n- **RHVoice**，根据 [GNU 通用公共许可证 v2.0](https:\u002F\u002Fraw.githubusercontent.com\u002FRHVoice\u002FRHVoice\u002Fmaster\u002FLICENSE.md) 发布\n- **ssplit-cpp**，根据 [Apache 许可证 2.0](https:\u002F\u002Fgithub.com\u002Fugermann\u002Fssplit-cpp\u002Fraw\u002Fmaster\u002FLICENSE.md) 发布\n- **espeak-ng**，根据 [GNU 通用公共许可证 v3.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fespeak-ng\u002Fespeak-ng\u002Fmaster\u002FCOPYING) 发布\n- **bergamot-translator**，根据 [Mozilla 公共许可证 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fbrowsermt\u002Fbergamot-translator\u002Fmain\u002FLICENSE) 发布\n- **Rubber Band Library**，根据 [GNU 通用公共许可证（版本 2 或更高）](https:\u002F\u002Fbreakfastquay.com\u002Frubberband\u002Flicense.html) 发布\n- **simdjson**，根据 [Apache 许可证 2.0](https:\u002F\u002Fgithub.com\u002Fsimdjson\u002Fsimdjson\u002Fraw\u002Fmaster\u002FLICENSE) 发布\n- **Nlohmann JSON**，根据 [MIT 许可证](https:\u002F\u002Fjson.nlohmann.me\u002Fhome\u002Flicense\u002F) 发布\n- **uroman**，根据 [此许可证](https:\u002F\u002Fgithub.com\u002Fisi-nlp\u002Furoman\u002Fraw\u002Fmaster\u002FLICENSE.txt) 发布\n- **astrunc**，根据 [MIT 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002FJoke-Shi\u002Fastrunc\u002Fmaster\u002FLICENSE) 发布\n- **FFmpeg**，根据 [GNU 较小通用公共许可证 2.1 版或更高](https:\u002F\u002Fgit.ffmpeg.org\u002Fgitweb\u002Fffmpeg.git\u002Fblob_plain\u002FHEAD:\u002FLICENSE.md) 发布\n- **LAME**，根据 LGPL 发布\n- **Vorbis**，根据 [此许可证](https:\u002F\u002Fgitlab.xiph.org\u002Fxiph\u002Fvorbis\u002F-\u002Fraw\u002Fmaster\u002FCOPYING?ref_type=heads) 发布\n- **TagLib**，根据 [GNU 较小通用公共许可证 (LGPL)](https:\u002F\u002Fraw.githubusercontent.com\u002Ftaglib\u002Ftaglib\u002Fmaster\u002FCOPYING.LGPL) 和 [Mozilla 公共许可证 (MPL)](https:\u002F\u002Fraw.githubusercontent.com\u002Ftaglib\u002Ftaglib\u002Fmaster\u002FCOPYING.MPL) 发布\n- **libnumbertext**，根据 [BSD 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002FNumbertext\u002Flibnumbertext\u002Fmaster\u002FCOPYING) 发布\n- **KDBusAddons**，根据 [LGPL 许可证](https:\u002F\u002Finvent.kde.org\u002Fframeworks\u002Fkdbusaddons\u002F-\u002Ftree\u002Fmaster\u002FLICENSES?ref_type=heads) 发布\n- **QHotkey**，根据 [BSD-3-Clause 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002FSkycoder42\u002FQHotkey\u002Fmaster\u002FLICENSE) 发布\n- **faster-whisper**，根据 [MIT 许可证](https:\u002F\u002Fgithub.com\u002Fguillaumekln\u002Ffaster-whisper\u002Fraw\u002Fmaster\u002FLICENSE) 发布\n- **Mimic 3**，根据 [AGPL-3.0 许可证](https:\u002F\u002Fgithub.com\u002FMycroftAI\u002Fmimic3\u002Fraw\u002Fmaster\u002FLICENSE) 发布\n- **Unikud**，根据 [MIT 许可证](https:\u002F\u002Fgithub.com\u002Fmorrisalp\u002Funikud\u002Fraw\u002Fmain\u002FLICENSE) 发布\n- **april-asr**，根据 [GNU 通用公共许可证 v3.0](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr\u002Fraw\u002Fmain\u002FCOPYING) 发布\n- **libopus**，根据 [此许可证](https:\u002F\u002Fgitlab.xiph.org\u002Fxiph\u002Fopus\u002F-\u002Fraw\u002Fmaster\u002FCOPYING) 发布\n- **html2md**，根据 [MIT 许可证](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT) 发布\n- **maddy**，根据 [MIT 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002Fprogsource\u002Fmaddy\u002Fmaster\u002FLICENSE) 发布\n- **WhisperSpeech**，根据 [MIT 许可证](https:\u002F\u002Fraw.githubusercontent.com\u002Fcollabora\u002FWhisperSpeech\u002Fmain\u002FLICENSE) 发布\n- **Kokoro**，根据 [Apache 许可证 2.0](https:\u002F\u002Fgithub.com\u002Fhexgrad\u002Fkokoro\u002Fraw\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE) 发布\n- **Parler-TTS**，根据 [Apache 许可证 2.0](https:\u002F\u002Fraw.githubusercontent.com\u002Fhuggingface\u002Fparler-tts\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE) 发布\n- **F5-TTS**，根据 [MIT 许可证](https:\u002F\u002Fgithub.com\u002FSWivid\u002FF5-TTS\u002Fraw\u002Frefs\u002Fheads\u002Fmain\u002FLICENSE) 发布\n\n`nonbreaking_prefixes` 目录中的文件是从\n[mosesdecoder](https:\u002F\u002Fgithub.com\u002Fmoses-smt\u002Fmosesdecoder) 项目复制而来，并根据\n[GNU 较小通用公共许可证 v2.1](https:\u002F\u002Fgithub.com\u002Fmoses-smt\u002Fmosesdecoder\u002Fraw\u002Fmaster\u002FCOPYING) 分发。","# Speech Note 快速上手指南\n\nSpeech Note 是一款专为 Linux 桌面和 Sailfish OS 设计的离线笔记应用，支持语音转文字（STT）、文字转语音（TTS）及机器翻译。所有数据处理均在本地完成，无需联网，充分保障隐私。\n\n## 环境准备\n\n### 系统要求\n- **操作系统**：主流 Linux 发行版（如 Ubuntu, Fedora, Debian, Arch 等）或 Sailfish OS。\n- **架构**：x86_64 或 aarch64（ARM64）。\n- **依赖管理**：推荐安装 `Flatpak` 以简化依赖管理和沙箱运行。\n\n### 前置依赖\n若不使用 Flatpak 而选择源码编译，需安装以下开发库（以 Ubuntu\u002FDebian 为例）：\n```bash\nsudo apt-get install git cmake build-essential libssl-dev libasound2-dev libpulse-dev\n# 其他特定引擎依赖请参考源码构建章节\n```\n*注：对于普通用户，强烈建议使用 Flatpak 版本，无需手动配置复杂的环境依赖。*\n\n## 安装步骤\n\n### 方法一：通过 Flathub 安装（推荐）\n这是最简单且稳定的安装方式，自动处理所有运行时依赖。\n\n1. 确保已安装 Flatpak：\n ```bash\n # Ubuntu\u002FDebian\n sudo apt install flatpak\n # Fedora\n sudo dnf install flatpak\n ```\n\n2. 添加 Flathub 仓库（如果尚未添加）：\n ```bash\n flatpak remote-add --if-not-exists flathub https:\u002F\u002Fflathub.org\u002Frepo\u002Fflathub.flatpakrepo\n ```\n\n3. 安装 Speech Note：\n ```bash\n flatpak install flathub net.mkiol.SpeechNote\n ```\n\n### 方法二：国内加速安装（可选）\n如果访问 Flathub 速度较慢，可配置国内镜像源（如上海交通大学镜像源）：\n\n```bash\nflatpak remote-modify flathub --url=https:\u002F\u002Fmirror.sjtu.edu.cn\u002Fflathub\nflatpak install flathub net.mkiol.SpeechNote\n```\n\n## 基本使用\n\n### 1. 启动应用\n安装完成后，可通过应用菜单搜索 \"Speech Note\" 启动，或使用命令行：\n```bash\nflatpak run net.mkiol.SpeechNote\n```\n\n### 2. 下载语言模型\n首次使用时，应用内不包含具体的模型文件，需通过内置浏览器下载：\n- 打开应用，进入 **Settings (设置)** 或 **Models (模型)** 面板。\n- 在列表中选择你需要的语言（如 Chinese, English 等）。\n- 勾选所需的引擎类型：\n - **STT**: 语音识别（推荐 `Whisper` 或 `Vosk`）。\n - **TTS**: 语音合成（推荐 `Piper` 或 `espeak`）。\n - **MT**: 机器翻译（`Bergamot`）。\n- 点击 **Download** 按钮，模型将自动下载并存储到本地。\n\n### 3. 核心功能示例\n\n#### 语音记笔记 (Speech to Text)\n1. 在编辑器界面，点击麦克风图标。\n2. 选择已下载的 STT 模型（例如 `whisper-large-v3`）。\n3. 对着麦克风说话，语音将实时转换为文字并填入笔记区域。\n\n#### 朗读笔记 (Text to Speech)\n1. 在笔记中输入或粘贴一段文字。\n2. 选中该段文字。\n3. 点击扬声器图标，选择已下载的 TTS 模型（例如 `piper-zh_CN`）。\n4. 应用将离线朗读所选内容。\n\n#### 离线翻译 (Machine Translation)\n1. 输入源语言文本。\n2. 点击翻译图标，选择目标语言。\n3. 系统将利用本地 Bergamot 模型立即生成翻译结果，全程无网络请求。\n\n> **提示**：所有操作均在断网环境下可用，一旦模型下载完成，无需再次连接互联网即可使用全部功能。","一位常驻德国的中国籍嵌入式工程师，需要在无网络的实验室环境中，快速记录德语设备调试语音日志并转化为中文技术文档。\n\n### 没有 dsnote 时\n- **网络依赖严重**：实验室屏蔽外网，无法使用在线语音转文字或翻译服务，只能手动听写，效率极低。\n- **隐私合规风险**：若用手机录音后联网处理，敏感的设备调试数据可能泄露，违反公司数据安全规定。\n- **多工具切换繁琐**：需分别使用录音笔、OCR 软件和在线翻译网页，数据分散且格式整理耗时。\n- **语言支持受限**：普通离线工具难以同时高质量支持“德语识别”与“中德互译”，专业术语识别率低。\n\n### 使用 dsnote 后\n- **完全离线作业**：dsnote 利用本地 Whisper 或 Vosk 模型，在无网环境下实时将德语语音转为文本，即说即记。\n- **数据本地闭环**：所有语音处理、翻译及朗读均在本地完成，无需上传云端，彻底消除数据泄露隐患。\n- **一站式工作流**：在 dsnote 单一界面内完成“德语语音输入 -> 自动翻译为中文 -> 中文语音复核”，笔记自动生成。\n- **专业模型适配**：可下载针对性的德语工程模型和 Bergamot 离线翻译包，显著提升专业术语的转写与翻译准确度。\n\ndsnote 通过全链路离线智能处理，让敏感环境下的跨语言技术笔记记录变得安全、高效且流畅。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmkiol_dsnote_877d9a58.png","mkiol","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fmkiol_03138c34.jpg",null,"https:\u002F\u002Fgithub.com\u002Fmkiol",[80,84,88,92,96],{"name":81,"color":82,"percentage":83},"C++","#f34b7d",85.5,{"name":85,"color":86,"percentage":87},"QML","#44a51c",11.9,{"name":89,"color":90,"percentage":91},"CMake","#DA3434",2.3,{"name":93,"color":94,"percentage":95},"Shell","#89e051",0.3,{"name":97,"color":98,"percentage":99},"C","#555555",0,1412,61,"2026-04-04T18:32:35","MPL-2.0","Linux, Sailfish OS","未说明 (支持多种引擎，部分如 Faster Whisper 可能受益于 GPU，但 README 未明确具体显卡型号、显存或 CUDA 版本要求)","未说明",{"notes":108,"python":106,"dependencies":109},"该工具主要通过 Flatpak 分发，支持 Linux 桌面和 Sailfish OS。所有语音和文本处理均在本地离线运行，无需网络连接。安装包不包含模型文件，需在首次使用时通过内置的图形化模型浏览器下载所需的语言模型（涵盖语音转文字、文字转语音和机器翻译）。支持多种后端引擎（如 Whisper, Vosk, Piper 等），用户可根据需求在应用内选择下载特定模型。",[110,111,112,113,114,115,116,117],"Coqui STT","Vosk","whisper.cpp","Faster Whisper","espeak-ng","Piper","Bergamot Translator","Flatpak (主要分发格式)",[38,15],[120,121,122,123,124,125,126,127,128,129,130,131,132,133,134],"asr","sailfishos","stt","tts","flatpak-applications","linux-desktop","nmt","offline","translator","machine-translation","speech-recognition","speech-synthesis","speech-to-text","text-to-speech","translation","2026-03-27T02:49:30.150509","2026-04-06T06:43:59.727376",[138,143,147,152,157,162,167],{"id":139,"question_zh":140,"answer_zh":141,"source_url":142},17297,"如何在 Flatpak 中启用测试版（Beta）仓库以体验新功能？","要在系统中启用 \"flathub-beta\" 仓库，可以运行以下命令：\n\nflatpak remote-add --if-not-exists flathub-beta https:\u002F\u002Fflathub.org\u002Fbeta-repo\u002Fflathub-beta.flatpakrepo\n\n启用后，即可安装或更新到 Beta 版本（例如 Speech Note 4.8.0 Beta），该版本通常包含最新的功能修复和新模型支持，但可能比正式版存在更多 Bug。","https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Fissues\u002F223",{"id":144,"question_zh":145,"answer_zh":146,"source_url":142},17298,"如何使用命令行将合成的语音导出为音频文件而不是直接播放？","可以使用 `--output-file` 参数配合 `start-reading-clipboard` 或 `start-reading-text` 动作来实现。例如，通过命令行调用时指定输出文件路径，程序会将合成后的语音保存为该文件，而不会通过扬声器播放。这适用于需要批量处理或保存语音文件的场景。",{"id":148,"question_zh":149,"answer_zh":150,"source_url":151},17299,"在 Wayland 环境下复制粘贴功能失效或需要手动切换快捷键怎么办？","在 Wayland 下，由于安全限制，应用无法直接模拟键盘操作进行粘贴。解决方案是改用剪贴板 DBus 服务（如 Plasma 的 `org.kde.klipper` 或通用的 `com.github.hluk.copyq`）或直接嵌入 `wl-clipboard` 二进制文件来处理剪贴板交互。\n\n维护者已在后续版本中修复了全局快捷键每次启动需手动切换的问题，并实现了对 wl-clipboard 的原生支持，同时保留了对 KDE Klipper 和 QClipboard 的降级兼容。建议升级到最新版本以获得最佳体验。","https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Fissues\u002F338",{"id":153,"question_zh":154,"answer_zh":155,"source_url":156},17300,"使用 Nvidia 专有驱动时应用启动崩溃如何解决？","部分用户在使用 Nvidia 专有驱动（特别是在 Debian Sid 等滚动发行版上）时遇到启动崩溃。根据社区反馈，执行干净的重新安装（clean install）通常能解决此问题。如果问题依旧，请检查 Flatpak 运行时环境是否正确加载了必要的图形库，或尝试更新显卡驱动及 Flatpak 运行时至最新版本。","https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Fissues\u002F153",{"id":158,"question_zh":159,"answer_zh":160,"source_url":161},17301,"使用 Coqui 模型进行文本转语音（TTS）时应用崩溃怎么办？","这是一个已知问题，特别是在 Linux\u002FUbuntu 搭配 Nvidia 显卡的环境下。该问题已在版本 v4.8.2 中修复。\n\n解决方法：请升级到 Speech Note v4.8.2 或更高版本。如果正式版尚未推送，可以先切换到 `flatpak-beta` 通道获取包含修复的版本：\n\nflatpak update --branch=beta net.mkiol.SpeechNote","https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Fissues\u002F299",{"id":163,"question_zh":164,"answer_zh":165,"source_url":166},17302,"在 openSUSE 上源码编译时遇到 FFmpeg、Vosk 或 OpenCL 相关的链接错误如何处理？","在 openSUSE 上构建时可能会遇到以下常见问题：\n1. **FFmpeg 链接错误**：若使用 `-DBUILD_FFMPEG=OFF` 报错，通常是 CMakeLists.txt 中的拼写错误导致，需应用相关补丁修复符号引用。\n2. **Vosk 头文件缺失**：若使用 `-DBUILD_VOSK=ON` 报错 `fst\u002Ftypes.h: No such file or directory`，尽管文件存在但编译器未找到，需应用 vosk.patch 修正包含路径。\n3. **OpenCL 库缺失**：若使用 `-DBUILD_WHISPERCPP_CLBLAST=ON` 报错找不到 OpenCL 库，是因为构建脚本只下载了头文件而未编译库。解决方法是预先安装系统的 OpenCL 开发包，或修改构建脚本确保先编译 OpenCL 库再编译 clblast。\n\n建议直接使用维护者发布的 openSUSE 安装包（如通过 Packman 仓库安装 v4.8.1+），以避免手动编译的复杂性。","https:\u002F\u002Fgithub.com\u002Fmkiol\u002Fdsnote\u002Fissues\u002F163",{"id":168,"question_zh":169,"answer_zh":170,"source_url":142},17303,"有哪些新推出的语音识别模型推荐，它们有什么特点？","近期引入了几个高质量的新模型：\n1. **CrisperWhisper**：专为 FasterWhisper 引擎设计，适用于英语和德语。它能逐字转录（verbatim），准确记录填充词、停顿、口吃等细节，适合需要精确时间戳和原始录音内容的场景。\n2. **KBLab Whisper**：瑞典国家图书馆发布的针对瑞典语微调的模型，相比常规 Whisper 模型准确率显著提升。\n3. **FUTO Whisper**：源自 FUTO 移动端键盘应用的新模型，优化了移动场景下的识别效果。\n\n此外，新版本支持利用现有笔记作为解码的初始上下文，有助于提高转录质量并减少“幻觉”现象。",[172,177,182,187,192,197,202,207,212,217,222,227,232,237,242,247,252,257,262,267],{"id":173,"version":174,"summary_zh":175,"released_at":176},99513,"v4.8.3","## Linux 桌面版\n\n更改：\n\n- 通用\n - 修复：当因错误无法下载许可证文件时，模型无法激活。\n- 语音转文本\n - 修复：在某些 CPU 上使用 _WhisperCpp_ 时，应用会崩溃。\n- 文本转语音\n - 修复：无法下载 Coqui XTTS 模型的许可证。\n- 翻译器\n - 修复：在某些 CPU 上使用翻译器时，应用会崩溃。","2025-08-15T15:12:36",{"id":178,"version":179,"summary_zh":180,"released_at":181},99514,"v4.8.2","## Linux 桌面\n\n变更：\n\n- 文本转语音\n - 新增阿根廷西班牙语、印地语、马拉雅拉姆语和尼泊尔语的 _Piper_ 语音\n - 修复：在某些平台上使用 _Coqui TTS_ 引擎会导致应用崩溃。\n - 修复：当文本转语音引擎生成损坏的音频文件时导致的崩溃\n- 语音转文本\n - 在 _Whisper_ 中启用的新语言：阿塞拜疆语、白俄罗斯语、卡纳达语、马拉雅拉姆语、泰米尔语\n- Flatpak\n - 将 _numba_ Python 包降级至 0.60.0 版本\n\n## Sailfish OS\n\n变更：\n\n- 文本转语音\n - 新增阿根廷西班牙语、印地语、马拉雅拉姆语和尼泊尔语的 _Piper_ 语音\n - 修复：当文本转语音引擎生成损坏的音频文件时导致的崩溃\n- 语音转文本\n - 在 _Whisper_ 中启用的新语言：阿塞拜疆语、白俄罗斯语、卡纳达语、马拉雅拉姆语、泰米尔语","2025-08-02T18:05:21",{"id":183,"version":184,"summary_zh":185,"released_at":186},99528,"v4.0.0","Changes:\r\n\r\n- Translator:\r\n - Support for offline translations.\r\n- Interface:\r\n - User interface redesign\r\n - Settings option to force specific interface style.\r\n - App translated to new languages: Dutch and Italian\r\n- Text to Speech:\r\n - All existing Piper models were updated.\r\n - New Piper voices for: English, Swedish, Turkish, Polish,\r\n German, Spanish, Finnish, French, Ukrainian, Russian,\r\n Swahili, Serbian, Romanian, Luxembourgish and Georgian\r\n - New RHVoice model for Slovak language","2023-08-07T15:11:14",{"id":188,"version":189,"summary_zh":190,"released_at":191},99529,"v3.1.5","## Changes in Linux Desktop version:\r\n\r\n- Text to Speech:\r\n - New Coqui voice for English: Jenny\r\n- Speech to Text:\r\n - Quicker decoding when using DeepSpeech\u002FCoqui models (especially on ARM CPU)\r\n\r\n## Changes in Sailfish OS version:\r\n\r\n- Speech to Text:\r\n - Quicker decoding when using DeepSpeech\u002FCoqui models\r\n - Re-enabled Swedish Vosk model","2023-07-07T12:30:01",{"id":193,"version":194,"summary_zh":195,"released_at":196},99515,"v4.8.1","## Linux 桌面版\n\n变更：\n\n- 翻译器\n - 修复：葡萄牙语、荷兰语、波斯语、挪威语和冰岛语的语言模型下载错误。\n - 更新了准确率更高的模型：德语到英语、荷兰语到英语、英语到乌克兰语、英语到匈牙利语、英语到加泰罗尼亚语、加泰罗尼亚语到英语、英语到立陶宛语、英语到拉脱维亚语、英语到斯洛文尼亚语、斯洛文尼亚语到英语、英语到斯洛伐克语、英语到俄语。\n - 新增模型：阿塞拜疆语到英语、白俄罗斯语到英语、孟加拉语到英语、古吉拉特语到英语、希伯来语到英语、印地语到英语、卡纳达语到英语、马拉雅拉姆语到英语、马来语到英语、阿尔巴尼亚语到英语、泰米尔语到英语。\n- 语音转文本\n - 新增德语的超大型 _Vosk_ 模型：_Tuda-DE Large_。\n- 文本转语音\n - _Coqui_ MMS 模型支持以下新语言：卡纳达语、马拉雅拉姆语、泰米尔语。\n- 用户界面\n - _Speech Note_ 已翻译为德语。 \n\n## Sailfish OS\n\n变更：\n\n- 翻译器\n - 修复：葡萄牙语、荷兰语、波斯语、挪威语和冰岛语的语言模型下载错误。\n - 更新了准确率更高的模型：德语到英语、荷兰语到英语、英语到乌克兰语、英语到匈牙利语、英语到加泰罗尼亚语、加泰罗尼亚语到英语、英语到立陶宛语、英语到拉脱维亚语、英语到斯洛文尼亚语、斯洛文尼亚语到英语、英语到斯洛伐克语、英语到俄语。\n - 新增模型：阿塞拜疆语到英语、白俄罗斯语到英语、孟加拉语到英语、古吉拉特语到英语、希伯来语到英语、印地语到英语、卡纳达语到英语、马拉雅拉姆语到英语、马来语到英语、阿尔巴尼亚语到英语、泰米尔语到英语。\n- 用户界面\n - _Speech Note_ 已翻译为德语。","2025-07-12T17:21:40",{"id":198,"version":199,"summary_zh":200,"released_at":201},99516,"v4.8.0","## Linux 桌面版\n\n全新功能视频演示：https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=ww6skKOOzZ8\n\n变更内容：\n\n- 通用\n - 在_规则_中支持区分大小写的匹配\n- 用户界面\n - _语音笔记_现已支持阿拉伯语、加泰罗尼亚语、西班牙语、土耳其语和加拿大法语。\n - 新增命令行选项和 D-Bus API，用于将合成语音导出为音频文件，而非直接播放。请将 `--output-file` 与 `start-reading-clipboard` 或 `start-reading-text` 操作配合使用。\n- 语音转文本\n - 为 _FasterWhisper_ 引擎新增 [_CrisperWhisper_](https:\u002F\u002Fgithub.com\u002Fnyrahealth\u002FCrisperWhisper) 模型。_CrisperWhisper_ 专为快速、精准且逐字逐句的语音识别而设计，并能提供准确的词级时间戳。与原生 Whisper 不同，后者倾向于省略口误并更注重意图化的转录风格，_CrisperWhisper_ 则旨在完整记录每一句口语内容，包括填充词、停顿、结巴以及重来的情况。目前，_CrisperWhisper_ 模型仅适用于英语和德语。\n - 为瑞典语新增 [_KBLab Whisper_](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002FKBLab\u002Fkb-whisper-67af9eafb24da903b63cc4aa) 模型。瑞典国家图书馆发布了基于其馆藏数据微调的 STT 模型，与常规 Whisper 模型相比，这些模型的准确性显著提升。\n - _FUTO_ Whisper 模型。这些是 FUTO 手机键盘应用中使用的新型模型。\n - 在解码过程中将现有笔记用作初始上下文。此举有望提升转录质量并减少“幻觉”问题。若发现转录质量下降，请关闭“将笔记用作上下文”选项。\n - 增加在处理过程中暂停监听的选项。此选项在“监听模式”设置为“始终开启”时尤为有用。默认情况下，即使正在处理一段音频数据，监听仍会继续进行。启用该选项后，可在处理期间暂时暂停监听。\n - 增加在开始和停止监听时播放提示音的选项。\n- 文本转语音\n - [_Kokoro_ TTS](https:\u002F\u002Fgithub.com\u002Fhexgrad\u002Fkokoro) 引擎。Kokoro 是一款体积小巧但功能强大的开源多语言 TTS 引擎。尽管训练数据不足 100 小时，其表现却令人印象深刻。Kokoro 语音现已支持英语、中文、日语、印地语、意大利语、法语、西班牙语和葡萄牙语。\n - [_F5-TTS_](https:\u002F\u002Fgithub.com\u002FSWivid\u002FF5-TTS) 引擎。F5-TTS 具备卓越的语音克隆能力。当前启用的模型支持英语和中文。F5-TTS 在 CUDA 加速下效果最佳，仅使用 CPU 处理时速度可能会非常慢。\n - [_Parler-TTS_](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fparler-tts) 引擎。Parler-TTS 能够根据给定说话者的特征（性别、音高、语速、表达方式等），生成高质量、自然流畅的语音。说话者的具体特征由文本描述（提示词）定义。要使用 Parler-TTS 模型，您需要配置一个 _","2025-06-20T16:37:54",{"id":203,"version":204,"summary_zh":205,"released_at":206},99517,"b4.8.0-4","语音笔记 4.8.0 测试版 4","2025-05-27T16:46:47",{"id":208,"version":209,"summary_zh":210,"released_at":211},99518,"v4.7.1","## Linux 桌面版\n\n更改：\n\n- 通用\n - 修复：当处理器不支持 AVX CPU 扩展时，应用程序无法启动。\n- 翻译器：\n - 新增模型：韩语到英语、日语到英语\n - 更新模型：中文到英语\n","2025-01-06T20:07:18",{"id":213,"version":214,"summary_zh":215,"released_at":216},99519,"v4.7.0","## Linux 桌面版\n\n更改：\n\n- 通用\n - 文本转换的_规则_，可在语音转文本之后或文本转语音之前应用。借助_规则_，您可以轻松灵活地纠正解码文本中的错误或发音不准的词语。\n - 新增在光标位置插入文本或替换当前笔记的模式。若希望在光标位置而非笔记末尾插入文本，请在设置中将_文本追加模式_选项更改为_在光标位置添加_。当启用_替换现有笔记_选项时，每次添加新文本都会覆盖当前笔记。\n - 用于与外部应用程序集成的DBus API\n- 用户界面：\n - Speech Note 已翻译为斯洛文尼亚语。\n - 系统托盘图标中的状态指示。使用系统托盘图标时，处理、监听等状态会通过动画托盘图标显示。\n - 模型浏览器中按类型分组的模型。为提升易用性，不再以包含所有类型模型的列表形式呈现，而是将模型按类型分别置于不同的标签页中。\n - _设置_中新增_通用_和_高级_两个标签页。\n - 命令行选项，用于打印可用或激活的模型ID。使用_--print-available-models_或_--print-active-model_可列出所有可用模型或当前激活的模型。\n - 命令行选项，用于打印应用程序的当前状态。使用_--print-state_即可查看当前状态。此选项在与外部程序或桌面小部件集成时非常有用。\n- 语音转文本：\n - _WhisperCpp_ 支持 Vulkan GPU 加速。Vulkan 加速可显著提升使用 Intel、AMD 或 NVIDIA 显卡时的语音转文本解码速度。与 OpenVINO、OpenCL 和 ROCm 相比，Vulkan 解码速度更快，但可能仍略逊于 CUDA。Vulkan 的最大优势在于无需安装任何 GPU 加速插件即可使用。\n - _WhisperCpp_ 和 _FasterWhisper_ 均新增了_Whisper Large Turbo_ 模型。_Turbo_ 是经过微调的剪枝版_Whisper Large-v3_，两者本质上完全相同，只是解码层的数量有所减少。因此，该模型的速度大幅提升，但质量会略有下降。_Turbo_ 模型不具备常规 Large 模型所具备的英译能力。\n - 简化了引擎配置选项。现在无需单独设置多个选项，只需选择一个配置文件，即可更改引擎的处理参数。共有三种配置文件可供选择：_最佳性能_、_最佳质量_和_自定义_。\n - 回声模式。处理完成后，解码后的文本将立即使用当前设置的文本转语音模型朗读出来。\n- 文本转语音：\n - 为拉脱维亚语新增_Piper_ 语音。\n - 新增_WhisperSpeech Small_ 模型，适用于英语、意大利语、德语、法语、西班牙语、荷兰语和葡萄牙语。\n- 翻译器：\n - 新增以下模型：英语到芬兰语、英语到土耳其语、英语到瑞典语、瑞典语到英语、英…","2024-12-29T09:42:24",{"id":218,"version":219,"summary_zh":220,"released_at":221},99520,"v4.6.1","## Linux 桌面\n\n变更：\n\n- 通用\n - 修复：当处理器不支持所需的 CPU 扩展时，应用程序无法启动。\n- 用户界面\n - 瑞典语翻译已更新。\n- 无障碍功能\n - 修复：特殊键盘按键未被支持为快捷键。例如：“收藏夹”、“启动邮件”、“刷新”、“主页”、“计算器”等更多按键……\n- 翻译器\n - 新增模型：英语到拉脱维亚语、英语到丹麦语、英语到克罗地亚语、英语到斯洛文尼亚语、印尼语到英语、罗马尼亚语到英语\n - 更新模型：英语到匈牙利语、捷克语到英语、希腊语到英语\n\n## Sailfish OS\n\n变更：\n\n- 用户界面\n - 瑞典语翻译已更新。\n- 翻译器\n - 新增模型：英语到拉脱维亚语、英语到丹麦语、英语到克罗地亚语、英语到斯洛文尼亚语、印尼语到英语、罗马尼亚语到英语\n - 更新模型：英语到匈牙利语、捷克语到英语、希腊语到英语","2024-08-17T08:47:30",{"id":223,"version":224,"summary_zh":225,"released_at":226},99521,"v4.6.0","## Linux 桌面版\n\n更改：\n\n- 用户界面\n - 语音笔记已翻译为挪威语。\n - 模型分组显示。提供多个子模型的模型（例如，提供不同声音的 TTS 模型）现在会以组的形式展示。这使得在模型浏览器中更容易找到所需的模型。\n- 语音转文本\n - 所有 Whisper 模型的名称已更改为 _WhisperCpp_，以更好地反映其背后的引擎。\n - STT 中的自动语言检测功能。要在 STT 过程中自动检测语言，请在语言列表中选择属于 _自动检测_ 类别的模型之一。\n - 引擎设置独立化。各引擎的配置现已在设置中分开。您可以分别为 _WhisperCpp_ 和 _FasterWhisper_ 设置参数。新增的配置选项包括：_同时线程数_、_束搜索宽度_、_音频上下文大小_、_使用 Flash Attention_。\n - 使用 _WhisperCpp_ 实现更快的解码。_WhisperCpp_ 新增了针对短句的优化功能，使 STT 的速度提升了一倍！\n - _WhisperCpp_ 引擎支持 OpenVINO 硬件加速。通过 OpenVINO，CPU 上的解码速度将大幅提升。如果您未使用 GPU 加速，建议在 _WhisperCpp_ 引擎设置中启用 OpenVINO。目前，OpenVINO 仅支持 CPU 加速。\n - 处理统计信息插入选项。新的设置选项允许在解码后将处理相关信息（如处理时间和音频长度）插入文本中。这对于比较不同模型、引擎及其参数的性能非常有用。\n- 文本转语音\n - 高级 TTS 处理的控制标签。通过控制标签，您可以动态调整合成文本的速度，或在句子之间添加静音。要使用控制标签，只需在文本中插入 _{speed: 0.5}_ 或 _{silence: 1s}_ 即可。为方便起见，您还可以通过文本上下文菜单中的 _插入控制标签_ 功能插入预定义的控制标签。\n - 威尔士语。新增威尔士语支持，由 _Piper_ 语音提供。\n - 西班牙语、意大利语和英语的新 _Piper_ 语音。\n - 斯洛伐克语和克罗地亚语的新 _RHVoice_ 语音。\n- 翻译器\n - 翻译器 UI 改进。_翻译_、_切换语言_ 和 _添加_ 按钮现位于两个文本框之间，使用起来更加便捷。\n - 对旧硬件的支持。此前，翻译器无法在不支持 CPU AVX 扩展的旧处理器上运行。现在已不再受此限制。\n - 新增模型：英语到立陶宛语、克罗地亚语到英语、拉脱维亚语到英语、丹麦语到英语、塞尔维亚语到英语、斯洛伐克语到英语、波斯尼亚语到英语、越南语到英语。\n - 更新模型：立陶宛语到英语、斯洛文尼亚语到英语、俄语到英语、乌克兰语到英语。\n- Flatpak\n - 新库：OpenVINO 版本 2024.1.0.15008。\n - whisper.cpp 更新至版本 1.6.2。\n - CTranslate2 更新至版本 4.3.1。\n\n所有内容的视频演示","2024-08-03T13:05:27",{"id":228,"version":229,"summary_zh":230,"released_at":231},99522,"v4.5.0","## Linux 桌面版\n\n更改：\n\n- 用户界面\n - 导入内嵌于视频文件中的字幕。如果您的视频文件包含一个或多个字幕流，您可以将选定的字幕导入记事本。\n - 支持更多字幕格式。您可以导入和导出 SRT、WebVTT 和 ASS 格式的字幕。\n - 统一的文件导入导出功能。文本、字幕、音频和视频文件均可通过统一的菜单栏选项进行导入或导出。\n - 设置选项：启用或禁用记住上次笔记的功能。若禁用该选项，重启应用后将无法恢复上次的笔记。\n - 设置选项：定义从文件导入笔记时的默认操作。您可以选择\u003Ci>询问是添加还是替换\u003C\u002Fi>、\u003Ci>添加到现有笔记\u003C\u002Fi>或\u003Ci>替换现有笔记\u003C\u002Fi>。\n - 增强的文本编辑器字体设置。您可以设置文本编辑器中使用的字体系列、样式和大小。\n - 文本修复选项：通过这些选项，您可以直接修正文本中的变音符号和标点符号。\n - 文本上下文菜单新增选项：\u003Ci>朗读选区\u003C\u002Fi>和\u003Ci>翻译选区\u003C\u002Fi>。使用鼠标右键可激活上下文菜单。\n - 新增文本追加方式：\u003Ci>空行之后\u003C\u002Fi>\n - 系统托盘菜单：用于切换当前使用的 STT\u002FTTS 模型\n - 音频输入设备采用更友好的名称\n - 简化了模型筛选功能。虽然灵活性有所降低，但更加易于理解和使用。\n - \u003Ci>语音笔记\u003C\u002Fi>现已支持乌克兰语和俄语。\n - 修复：取消操作会阻塞用户界面。\n\n- 语音转文本\n - 更新了英语版 [Distil](https:\u002F\u002Fhuggingface.co\u002Fdistil-whisper\u002Fdistil-large-v3) 模型：\u003Ci>Distil Large-v3\u003C\u002Fi>。新模型已应用于 Whisper 和 Faster Whisper 引擎。\n - 新训练的 Whisper 模型，分别适用于[斯洛文尼亚语](https:\u002F\u002Fhuggingface.co\u002Fsamolego\u002Fwhisper-small-slovenian)和[波兰语](https:\u002F\u002Fhuggingface.co\u002FAspik101\u002Fdistil-whisper-large-v3-pl)。\n - 修复：标点符号模型无法下载。\n\n- 文本转语音\n - [WhisperSpeech](https:\u002F\u002Fcollabora.github.io\u002FWhisperSpeech) 引擎，能够生成极具自然感的语音。该引擎提供英语和波兰语模型，所有模型均支持语音克隆。\n - 新的越南语语音克隆模型：[viXTTS](https:\u002F\u002Fhuggingface.co\u002Fcapleaf\u002FviXTTS)。该模型是基于出色的\u003Ci>Coqui XTTS\u003C\u002Fi>进一步优化而成。\n - 新增 Piper 语音，涵盖英语、波斯语、斯洛文尼亚语、土耳其语、法语和西班牙语。\n - 新增 RHVoice 捷克语语音。\n - 设置选项：启用或禁用语音与字幕时间戳同步。此功能可用于制作配音。\n - 将语音与现有文件中的音频混合。在导出为文件时，您可以将语音叠加到现有的媒体文件上，这在根据字幕制作配音时非常有用。\n - 上下文菜单新增选项：从光标位置开始朗读，或仅朗读选中文本。使用鼠标右键可激活上下文菜单。","2024-05-18T16:49:53",{"id":233,"version":234,"summary_zh":235,"released_at":236},99523,"v4.4.0","## Linux Desktop\r\n\r\nChanges:\r\n\r\n- Flatpak\r\n - Modular Flatpak package (Base package and Add-ons)\r\n - NVIDIA CUDA runtime update to version 12.2\r\n - AMD ROCm runtime update to version 5.6\r\n - PyTorch update to version 2.1.1\r\n- User Interface\r\n - Improvements to the model browser\r\n - Model filtering options\r\n - Setting option to minimize to the system tray\r\n - Setting option to enable\u002Fdisable text in desktop notifications\r\n- Speech to Text\r\n - Marathi language. New language is enabled with Whisper and Faster Whisper models.\r\n - New version of Faster Whisper Large model: 'FasterWhisper Large-v3'\r\n - 'Distil' versions of Faster Whisper models\r\n - Whisper and Faster Whisper enabled for Chinese-Cantonese language\r\n - Support for Speex audio codec in 'Transcribe a file'\r\n - Translate to English option for Whisper and Faster Whisper models\r\n - More effective GPU acceleration for Whisper models with AMD graphics cards\r\n - Subtitles generation (SRT format)\r\n - Support for multiple audio streams in a video file\r\n- Text to Speech\r\n - Marathi language. New language is enabled with Coqui MMS model.\r\n - Voice cloning with Coqui XTTS and YourTTS models.\r\n - Coqui XTTS models are enabled for: Arabic, Brazilian Portuguese, Chinese, Czech, Dutch, English, French, German, Hungarian, Italian, Japanese, Korean, Polish, Russian, Spanish and Turkish.\r\n - YourTTS model is enabled for: English, French and Brazilian Portuguese.\r\n - Voice samples creator\r\n - New voices for Serbian and Uzbek languages (RHVoice model)\r\n - GPU acceleration for Coqui models with AMD graphics cards (in Flatpak version)\r\n - Speech synchronized with subtitle timestamps\r\n- Translator\r\n - New model: Lithuanian to English\r\n - Option to force text cleaning before translation\r\n - Text formatting support\r\n - Translation progress indicator\r\n- Other\r\n - Setting option to override GPU version (AMD graphics cards)\r\n - Setting option to limit number of simultaneous CPU threads\r\n - Setting option to set Python libraries directory (in non-Flatpak version)\r\n\r\n## Sailfish OS\r\n\r\n- Speech to Text\r\n - Marathi language. New language is enabled with Whisper models.\r\n - Whisper enabled for Chinese-Cantonese language\r\n - Support for Speex audio codec in 'Transcribe a file'\r\n - Support for multiple audio streams in a video file\r\n- Text to Speech\r\n - New voices for Serbian and Uzbek languages (RHVoice model)\r\n- Translator\r\n - New model: Lithuanian to English\r\n - Translation progress indicator","2024-01-26T09:54:09",{"id":238,"version":239,"summary_zh":240,"released_at":241},99524,"v4.3.0","## Linux Desktop\r\n\r\nChanges:\r\n\r\n- Accessibility\r\n - Global keyboard shortcuts (X11 only)\r\n - Support for _Actions_\r\n- User Interface\r\n - Desktop notifications\r\n - _Speech speed_ control in the main app window\r\n - Opening files with Drag and Drop gesture\r\n - Fix: Application did not use native widgets on some platforms\r\n- Translator\r\n - New model: English to Hungarian\r\n- Speech to Text\r\n - New languages: Afrikaans, Gujarati, Hausa, Telugu, Tswana, Javanese, Hebrew\r\n - New engine: [Faster Whisper](https:\u002F\u002Fgithub.com\u002Fguillaumekln\u002Ffaster-whisper)\r\n - New engine: [April-ASR](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr). Models for: English, French and Polish.\r\n - Inserting text to any active window (X11 only)\r\n - Copy decoded text directly to the clipboard\r\n - Stop listening button\r\n - Support for Opus audio codec in _Transcribe a file_\r\n - More effective GPU acceleration for Whisper models (NVIDIA CUDA only)\r\n - New smaller and quicker Whisper models for English: [Distil-Whisper](https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fdistil-whisper\u002Ftree\u002Fmain)\r\n - New version of Whisper Large model: [Whisper Large-v3](https:\u002F\u002Fhuggingface.co\u002Fopenai\u002Fwhisper-large-v3)\r\n - Fix: CUDA acceleration for Whisper models did not work on NVIDIA video cards with Maxwell architecture\r\n- Text to Speech\r\n - New languages: Afrikaans, Gujarati, Hausa, Telugu, Tswana, Javanese, Hebrew\r\n - New engine: [Mimic 3](https:\u002F\u002Fmycroft.ai\u002Fmimic-3\u002F)\r\n - Reading text from the clipboard\r\n - New Piper voices: Arabic, English, Hungarian, Polish, Czech, German, Ukrainian, Vietnamese, Serbian, French, Spanish, Nepali\r\n - More steps in _Speech speed_ option\r\n - Diacritical marks restoration before speech synthesis for Arabic and Hebrew\r\n - Support for GPU acceleration for Coqui models (NVIDIA CUDA only)\r\n - Fix: Coqui Chinese MMS Hakka and MinNan voices were broken\r\n - Fix: Exporting to audio file was not possible when text was very long\r\n- Other\r\n - Setting option to disable support for certain graphic cards\r\n - Setting option _Clear cache on close_\r\n - Cache compression (Opus format instead of raw audio)\r\n - Detecting the availability of the optional features\r\n\r\n## Sailfish OS\r\n\r\nChanges:\r\n\r\n- Translator\r\n - New model: English to Hungarian\r\n- Speech to Text\r\n - New languages: Afrikaans, Gujarati, Hausa, Telugu, Tswana, Javanese, Hebrew\r\n - New engine: [April-ASR](https:\u002F\u002Fgithub.com\u002Fabb128\u002Fapril-asr). Models for: English, French and Polish.\r\n - Stop listening button\r\n - Support for Opus audio codec in _Transcribe a file_\r\n- Text to Speech\r\n - New Piper voices: Arabic, English, Hungarian, Polish, Czech, German, Ukrainian, Vietnamese, Serbian, French, Spanish, Nepali\r\n - More steps in _Speech speed_ option\r\n - Diacritical marks restoration before speech synthesis for Arabic\r\n - Fix: Exporting to audio file was not possible when text was very long\r\n- Other\r\n - Setting option _Clear cache on close_\r\n - Cache compression (Opus format instead of raw audio)","2023-11-13T08:53:52",{"id":243,"version":244,"summary_zh":245,"released_at":246},99525,"v4.2.1","## Linux Desktop\r\n\r\nChanges:\r\n\r\n- Speech to Text\r\n - Improved AMD GPU acceleration support for Whisper models","2023-09-29T18:01:24",{"id":248,"version":249,"summary_zh":250,"released_at":251},99526,"v4.2.0","## Linux Desktop\r\n\r\nChanges:\r\n\r\n- Translator\r\n - New models: Hungarian to English, Finnish to English\r\n- Speech to Text\r\n - Support for video files transcription\r\n - Option 'Audio source' to select preferred audio source\r\n - Whisper engine update and increase in performance.\r\n Processing time has been reduced by an average of 50%.\r\n - Improved Nvidia GPU acceleration support for Whisper models\r\n- Text to Speech\r\n - Save audio in compressed formats (MP3 or Ogg Vorbis).\r\n You can also save metadata tags to the audio file, such as track number, title, artist or album.\r\n - Pause option. You can pause or resume speech reading.\r\n - New MMS models: Hungarian, Catalan, German,\r\n Spanish, Romanian, Russian and Swedish\r\n - Update of RHVoice voice for Uzbek\r\n - Fix: Many Coqui models couldn't read the numbers or the reading wasn't correct.\r\n - Fix: Piper models could not be downloaded\r\n- User Interface\r\n - Menu options: 'Open a text file' and 'Save to a text file'\r\n - Command line option to open files\r\n - Improved UI colors when app is running under GNOME dark theme\r\n - Option 'Graphical style' to change Qt interface style\r\n\r\n## Sailfish OS\r\n\r\nChanges:\r\n\r\n- Translator\r\n - New models: Hungarian to English, Finnish to English\r\n- Speech to Text\r\n - Support for video files transcription. With 'Transcribe a file' menu option you can\r\n convert audio file or audio from video file to text.\r\n - Whisper engine update and increase in performance.\r\n Processing time has been reduced by an average of 15% (Xperia 10 III).\r\n- Text to Speech\r\n - Save audio in compressed formats (MP3 or Ogg Vorbis).\r\n You can also save metadata tags to the audio file, such as track number, title, artist or album.\r\n - Pause option. You can pause or resume speech reading.\r\n - Update of RHVoice voice for Uzbek\r\n - Fix: Piper models could not be downloaded\r\n- User Interface\r\n - Share to Speech Note. You can push text, audio or video content to Speech Note\r\n using share button in other apps (e.g. Notes, Gallery, Audio recorder, Browser).","2023-09-25T13:03:32",{"id":253,"version":254,"summary_zh":255,"released_at":256},99527,"v4.1.0","## Linux Desktop\r\n\r\nChanges:\r\n\r\n- Speech to Text:\r\n - Support for GPU acceleration for Whisper models\r\n - Fix: Whisper wasn't able to decode short speech sentences\r\n- Text to Speech:\r\n - Option 'Speech speed' to make synthesized speech slower or faster.\r\n - New models from Massively Multilingual Speech (MMS) project:\r\n Albanian, Amharic, Arabic, Basque, Bengali, Bulgarian, Chinese,\r\n Greek, Hindi, Icelandic, Indonesian, Kazakh, Korean, Latin,\r\n Latvian, Malay, Mongolian, Polish, Portuguese, Swahili, Tagalog,\r\n Tatar, Thai, Turkish, Uzbek, Vietnamese, Yoruba\r\n - New Piper voices: Czech, German, Hungarian, Portuguese, Slovak,\r\n English\r\n - Update of RHVoice voices for Slovak and Czech\r\n - New Coqui voices for Japanese, Turkish and Spanish\r\n - Fix: Splitting text into sentences was incorrect for: Georgian,\r\n Japanese, Bengali, Nepali, Hindi\r\n- Interface\r\n - Option to change font size in text editor\r\n\r\n## Sailfish OS\r\n\r\nChanges:\r\n\r\n- Speech to Text:\r\n - Remove of experimental 'Restore punctuation' option\r\n - Fix: Whisper wasn't able to decode short speech sentences\r\n- Text to Speech:\r\n - Option 'Speech speed' to make synthesized speech slower or faster.\r\n - New Piper voices: Czech, German, Hungarian, Portuguese, Slovak,\r\n English\r\n - Update of RHVoice voices for Slovak and Czech\r\n - Fix: Splitting text into sentences was incorrect for: Georgian,\r\n Japanese, Bengali, Nepali, Hindi","2023-08-23T14:53:10",{"id":258,"version":259,"summary_zh":260,"released_at":261},99530,"v3.1.4.1","## Changes in Linux Desktop version:\r\n\r\n- Interface:\r\n - Option to show recent changes (About -> Changes)\r\n - French translation update (Many thanks to @LAfricain)\r\n- Text to Speech:\r\n - New Piper model for Chinese\r\n - New RHVoice model for Uzbek (Beta)\r\n - Updated RHVoice models for Ukrainian\r\n - Piper and RHVoice engines updated to most recent versions\r\n- Speech to Text:\r\n - Whisper 'Large' models enabled for all languages\r\n - Whisper supported on older CPUs (i.e. without AVX\u002FAVX2 extensions)\r\n - Whisper engine update (20% performance improvement, 50% less memory)\r\n\r\n## Changes in Sailfish OS version:\r\n\r\n- Interface:\r\n - French translation update (Many thanks to @LAfricain)\r\n- Text to Speech:\r\n - New Piper model for Chinese\r\n - New RHVoice model for Uzbek (Beta)\r\n - Updated RHVoice models for Ukrainian\r\n - Piper and RHVoice engines updated to most recent versions\r\n- Speech to Text:\r\n - Whisper 'Small' models enabled for all languages\r\n - Whisper fine-tuned 'Small' models for: Croatian, Czech, Hungarian, Slovak and Romanian\r\n - Whisper engine update (test on Xperia 10 III: 20% performance improvement, 50% less memory)","2023-07-04T08:28:21",{"id":263,"version":264,"summary_zh":265,"released_at":266},99531,"v3.1.3","Changes:\r\n\r\n- New Piper Text-to-Speech models for: Icelandic, Swedish, Russian\r\n- Whisper 'fine-tuned' Speech-To-Text models for: Czech, Slovak, Slovenian, Romanian, Russian, Hungarian, Polish\r\n- Whisper models enabled also for: Amharic, Arabic, Bengali, Danish, Estonian, Basque, Persian, Hindi, Croatian, Hungarian, Icelandic, Georgian, Kazakh, Korean, Lithuanian, Latvian, Mongolian, Maltese, Nepali, Romanian, Slovak, Slovenian, Albanian, Swahili, Tagalog, Tatar, Uzbek, Yoruba","2023-06-24T17:20:06",{"id":268,"version":269,"summary_zh":270,"released_at":271},99532,"v3.1.1","Changes:\r\n\r\n- Option to save speech to audio file\r\n- New STT DeepSpeech model for Latvian language\r\n- Linux Desktop UI (Flatpak release on flathub.org)\r\n- Coqui TTS models for many languages (only in x86_64 Flatpak version)","2023-06-17T15:09:25"]