[{"data":1,"prerenderedAt":179},["ShallowReactive",2],{"repo-kittenml-kittentts-en":3,"i18n-kittenml-kittentts-en":176},{"mirrorPool":4,"detail":13,"relatedProjects":69},[5,9],{"key":6,"label":7,"base":8,"type":6},"github","mirror.github","https:\u002F\u002Fgithub.com\u002F",{"key":10,"label":11,"base":12,"type":10},"auto-proxy","mirror.autoProxy","https:\u002F\u002Fgithub.xxlab.tech\u002F",{"owner":14,"repo":15,"repoId":16,"repoUrl":17,"starCount":18,"language":19,"description":20,"languages":21,"screenshots":23,"homepage":24,"pushedAt":25,"ownerAvatar":26,"ownerName":24,"ownerBio":24,"lastSyncTime":27,"lastReleaseTime":28,"tags":29,"license":30,"category":31,"subgroup":32,"releases":33},"KittenML","KittenTTS",1032248449,"https:\u002F\u002Fgithub.com\u002FKittenML\u002FKittenTTS",16194,"Python","Open-source State-of-the-art TTS model which runs on a CPU 😻 ",{"Python":22},100,[],"","2026-10-16","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F224667585?v=4","2026-10-20T10:23:40.907347+00:00","2026-02-24T20:14:42Z",[],"Apache-2.0","ai","语音转文字工具",[34,48,58],{"tag":35,"publishAt":28,"semVer":36,"prerelease":40,"assets":41},"0.8.1",[37,38,39],0,8,1,false,[42],{"fileName":43,"sizeMb":44,"sha256":45,"downloadCount":46,"originUrl":47},"kittentts-0.8.1-py3-none-any.whl",0.02,"482a436c4f1f3192153710376e459ff3689517ebcda7c2b051e2fd4187b41851",1869400,"https:\u002F\u002Fgithub.com\u002FKittenML\u002FKittenTTS\u002Freleases\u002Fdownload\u002F0.8.1\u002Fkittentts-0.8.1-py3-none-any.whl",{"tag":49,"publishAt":50,"semVer":51,"prerelease":40,"assets":52},"0.8","2026-02-19T04:25:24Z",[37,37,37],[53],{"fileName":54,"sizeMb":44,"sha256":55,"downloadCount":56,"originUrl":57},"kittentts-0.8.0-py3-none-any.whl","cba97464da2ec56384b9fc03004449dadbe6d9abd16aa89e98953027854fd7d9",16048,"https:\u002F\u002Fgithub.com\u002FKittenML\u002FKittenTTS\u002Freleases\u002Fdownload\u002F0.8\u002Fkittentts-0.8.0-py3-none-any.whl",{"tag":59,"publishAt":60,"semVer":61,"prerelease":40,"assets":62},"0.1","2025-08-05T03:32:23Z",[37,37,37],[63],{"fileName":64,"sizeMb":65,"sha256":66,"downloadCount":67,"originUrl":68},"kittentts-0.1.0-py3-none-any.whl",0.01,"5a498422ec275f5df41b87592c6c482e637230fd563039197244d3966d1ce793",230220,"https:\u002F\u002Fgithub.com\u002FKittenML\u002FKittenTTS\u002Freleases\u002Fdownload\u002F0.1\u002Fkittentts-0.1.0-py3-none-any.whl",[70,91,110,132,147,163],{"owner":71,"repo":72,"repoId":73,"category":74,"subgroup":32,"language":75,"tags":76,"starCount":81,"ownerUid":82,"latestTag":83,"platforms":84,"lastReleaseTime":88,"downloadCount":89,"description":90},"cjpais","Handy",931888694,"tools","Rust",[77,78,79,80],"accessibility","cross-platform","speech-to-text","tauri-v2",32801,1559480,"v0.9.8",[85,86,87],"windows","mac","linux","2026-10-03T02:42:15Z",13374,"A free, open-source, and extensible speech-to-text application that works completely offline, supporting Windows, macOS, and Linux.",{"owner":92,"repo":93,"repoId":94,"category":31,"subgroup":32,"language":95,"tags":96,"starCount":102,"ownerUid":103,"latestTag":104,"platforms":105,"lastReleaseTime":107,"downloadCount":108,"description":109},"altic-dev","FluidVoice",1061327311,"Swift",[31,97,98,99,100,101],"dictation","ios","llama-cpp","macos","swift",12444,224717674,"v1.6.13-beta.1",[86,106],"other","2026-10-25T09:45:47Z",201,"The fastest macOS dictation app with on-device AI enhancement, a local open-source alternative to Wispr Flow.",{"owner":111,"repo":112,"repoId":113,"category":74,"subgroup":32,"language":114,"tags":115,"starCount":125,"ownerUid":126,"latestTag":127,"platforms":128,"lastReleaseTime":129,"downloadCount":130,"description":131},"OpenWhispr","openwhispr",1004647857,"JavaScript",[31,116,78,117,118,87,100,119,120,121,122,79,123,124,85],"anthropic","gemini","groq","nvidia","open-source","openai","parakeet","transcribe","whisper",9741,254803777,"v1.10.4",[85,86,87],"2026-10-22T08:21:23Z",4473,"A privacy-first cross-platform voice-to-text app with local (NVIDIA Parakeet\u002FWhisper) and cloud models, featuring meeting transcription and AI agent interaction.",{"owner":133,"repo":134,"repoId":135,"category":31,"subgroup":32,"language":136,"tags":137,"starCount":140,"ownerUid":141,"latestTag":142,"platforms":143,"lastReleaseTime":144,"downloadCount":145,"description":146},"thewh1teagle","vibe",740293013,"TypeScript",[31,78,138,121,139,123,124],"desktop","rust",7743,61390950,"v3.2.2",[85,86,87],"2026-09-05T13:24:57Z",62026,"A privacy-oriented, local AI transcription app using OpenAI's Whisper model that transcribes audio and video on your device.",{"owner":148,"repo":148,"repoId":149,"category":74,"subgroup":32,"language":150,"tags":151,"starCount":156,"ownerUid":157,"latestTag":158,"platforms":159,"lastReleaseTime":160,"downloadCount":161,"description":162},"espeak-ng",47648456,"C",[152,153,148,154,155],"android","espeak","speech-synthesis","text-to-speech",6910,16214005,"1.52.0",[85,152],"2024-12-12T13:47:27Z",291601,"A compact open-source speech synthesizer (TTS) supporting over 100 languages and accents, using formant synthesis method, available as a command-line tool, shared library, and SAPI5 version.",{"owner":164,"repo":165,"repoId":166,"category":74,"subgroup":32,"language":95,"tags":167,"starCount":169,"ownerUid":170,"latestTag":171,"platforms":172,"lastReleaseTime":173,"downloadCount":174,"description":175},"Beingpax","VoiceInk",875661263,[100,168,101],"macos-app",6720,101010368,"v2.22",[86],"2026-10-01T09:16:53Z",534,"The best open-source alternative to Superwhisper and Wispr Flow for macOS — a no-subscription voice-to-text app with local offline models, smart context awareness, global shortcuts, and an AI assistant mode.",{"description":177,"readmeHtml":178},"KittenTTS is an open-source text-to-speech library (Python). Its flagship KittenTTS 2 clones a voice from about 5 seconds of audio, controls expression, speaks 20 languages, and runs in real time on a CPU.","\u003Ch2 id=\"project-overview\">Project Overview\u003C\u002Fh2>\u003Cp>KittenTTS is an open-source text-to-speech library. The flagship model, KittenTTS 2, is a 1.7B-parameter 1-bit speech language model that clones any speaker from about 5 seconds of reference audio and supports emotion tags and emphasis control. It also keeps lightweight ONNX models from 15M to 80M parameters that run on a CPU with no GPU, synthesizing 24 kHz speech in real time.\u003C\u002Fp>\n\u003Cblockquote>\n\u003Cp>Voice cloning and lightweight deployment in one library that speaks in real time on a CPU.\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Item\u003C\u002Fth>\n\u003Cth>Value\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>Platform\u003C\u002Ftd>\n\u003Ctd>Linux, macOS, Windows (CPU or GPU)\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Language\u003C\u002Ftd>\n\u003Ctd>Python\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>License\u003C\u002Ftd>\n\u003Ctd>Apache-2.0\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Output rate\u003C\u002Ftd>\n\u003Ctd>24 kHz\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\n\u003Ch2 id=\"key-features\">Key Features\u003C\u002Fh2>\u003Cul>\n\u003Cli>\u003Cstrong>Voice cloning\u003C\u002Fstrong>: replicate any speaker from 5-30 seconds of audio, no fine-tuning\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Built-in voices\u003C\u002Fstrong>: 47 presets, including 9 non-English voices\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Multilingual\u003C\u002Fstrong>: 20 languages including Chinese, English, Japanese, Korean, French, and German\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Expression control\u003C\u002Fstrong>: emotion tags, event tags, and emphasis brackets\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Decoding presets\u003C\u002Fstrong>: trade stability against expressiveness per request\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Long-form text\u003C\u002Fstrong>: sentence-aware chunking for article-level reading\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Text preprocessing\u003C\u002Fstrong>: numbers, currencies, dates, and abbreviations expanded automatically\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Lightweight models\u003C\u002Fstrong>: 15M-80M ONNX models run without a GPU\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2 id=\"target-audience\">Target Audience\u003C\u002Fh2>\u003Cul>\n\u003Cli>Developers adding speech narration to apps\u003C\u002Fli>\n\u003Cli>Content creators cloning specific voices\u003C\u002Fli>\n\u003Cli>Engineers synthesizing speech on low-end or edge devices\u003C\u002Fli>\n\u003Cli>Teams producing audiobooks and voiceovers\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2 id=\"quick-start\">Quick Start\u003C\u002Fh2>\u003Col>\n\u003Cli>Install:\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cdiv class=\"code-block\">\u003Cpre>\u003Ccode class=\"hljs language-bash\">pip install kittenml\u003C\u002Fcode>\u003C\u002Fpre>\u003C\u002Fdiv>\u003Col start=\"2\">\n\u003Cli>Generate speech:\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cdiv class=\"code-block\">\u003Cpre>\u003Ccode class=\"hljs language-python\">from kittenml import KittenTTS\n\nm = KittenTTS('KittenML\u002Fkitten-tts-2')\naudio = m.generate('Hello, world', voice='Bruno')\u003C\u002Fcode>\u003C\u002Fpre>\u003C\u002Fdiv>\u003Col start=\"3\">\n\u003Cli>To clone a voice, pass reference instead of voice with your recording:\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cdiv class=\"code-block\">\u003Cpre>\u003Ccode class=\"hljs language-python\">audio = m.generate('This is a cloned voice', reference='my_voice.wav')\u003C\u002Fcode>\u003C\u002Fpre>\u003C\u002Fdiv>\u003Col start=\"4\">\n\u003Cli>Lightweight models load through the same KittenTTS constructor with no extra install\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Ch2 id=\"core-advantages\">Core Advantages\u003C\u002Fh2>\u003Cul>\n\u003Cli>\u003Cstrong>No GPU needed\u003C\u002Fstrong>: lightweight models run on CPU, and the flagship also has a CPU path\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Low cloning barrier\u003C\u002Fstrong>: a few seconds of audio replicates a voice without training\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Minimal install\u003C\u002Fstrong>: one package, weights download on demand, no Hugging Face login\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2 id=\"system-requirements\">System Requirements\u003C\u002Fh2>\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Item\u003C\u002Fth>\n\u003Cth>Requirement\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>OS\u003C\u002Ftd>\n\u003Ctd>Linux, macOS, Windows\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Runtime\u003C\u002Ftd>\n\u003Ctd>Python 3.9+\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Hardware\u003C\u002Ftd>\n\u003Ctd>GPU with about 8 GB VRAM, or a CPU with about 6 GB RAM\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\n\u003Ch2 id=\"notes\">Notes\u003C\u002Fh2>\u003Cul>\n\u003Cli>\u003Cstrong>Cloning input\u003C\u002Fstrong>: reference audio should be 5-30 seconds of a single speaker; mixed voices reduce similarity\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Weight choice\u003C\u002Fstrong>: default weights are 947 MiB lossless; \u003Ccode>weights=emb4\u003C\u002Fcode> halves the size at a small quality cost\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Legacy voices\u003C\u002Fstrong>: the eight voices from 0.8 remain in the new model, so older code keeps working\u003C\u002Fli>\n\u003C\u002Ful>\n",1793197885705]