#whisper — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #whisper, aggregated by home.social.
-
Étude du consortium ARIANE sur la transcription via IA générative avec le LLM Whisper : « Whisper : transcription automatique de sources sonores » https://hal.science/hal-05706209 #Whisper #LLM #ESR
-
LLM 改变了很多，比如 Agent 概念上半年爆火后，大厂们都争前恐后推出 CLI 产品，知乎，微信，网易云等等，这在几年前是完全无法想象的，那时想让 Linux 用上微信都很难，需要折腾一大堆 Wine 的组件，有时还会面临封号的风险。反观现在拥抱全平台的吃相，让人忍俊不禁。
但真正改变的，其实很有限，因为爆火的原因不是厂商良心，拥抱开源，而是发现这个市场有的赚，本质还是逐利的，Agent 让厂商发现开放生态有的赚。工具功能很全，很方便，但我却更加担心了，工具会一直免费下去吗？我猜不会，等到官方的生态建立起来，我想就是回血的时候了。
值得高兴的是，这些工具还很烂，如果真的要用，还太早，比如 https://www.npmjs.com/package/@music163/ncm-cli
我无法相信官方的 TUI 居然吹能播放一半不到的歌，真是充满恶意的设计
-
LLM 改变了很多，比如 Agent 概念上半年爆火后，大厂们都争前恐后推出 CLI 产品，知乎，微信，网易云等等，这在几年前是完全无法想象的，那时想让 Linux 用上微信都很难，需要折腾一大堆 Wine 的组件，有时还会面临封号的风险。反观现在拥抱全平台的吃相，让人忍俊不禁。
但真正改变的，其实很有限，因为爆火的原因不是厂商良心，拥抱开源，而是发现这个市场有的赚，本质还是逐利的，Agent 让厂商发现开放生态有的赚。工具功能很全，很方便，但我却更加担心了，工具会一直免费下去吗？我猜不会，等到官方的生态建立起来，我想就是回血的时候了。
值得高兴的是，这些工具还很烂，如果真的要用，还太早，比如 https://www.npmjs.com/package/@music163/ncm-cli
我无法相信官方的 TUI 居然吹能播放一半不到的歌，真是充满恶意的设计
-
LLM 改变了很多，比如 Agent 概念上半年爆火后，大厂们都争前恐后推出 CLI 产品，知乎，微信，网易云等等，这在几年前是完全无法想象的，那时想让 Linux 用上微信都很难，需要折腾一大堆 Wine 的组件，有时还会面临封号的风险。反观现在拥抱全平台的吃相，让人忍俊不禁。
但真正改变的，其实很有限，因为爆火的原因不是厂商良心，拥抱开源，而是发现这个市场有的赚，本质还是逐利的，Agent 让厂商发现开放生态有的赚。工具功能很全，很方便，但我却更加担心了，工具会一直免费下去吗？我猜不会，等到官方的生态建立起来，我想就是回血的时候了。
值得高兴的是，这些工具还很烂，如果真的要用，还太早，比如 https://www.npmjs.com/package/@music163/ncm-cli
我无法相信官方的 TUI 居然吹能播放一半不到的歌，真是充满恶意的设计
-
LLM 改变了很多，比如 Agent 概念上半年爆火后，大厂们都争前恐后推出 CLI 产品，知乎，微信，网易云等等，这在几年前是完全无法想象的，那时想让 Linux 用上微信都很难，需要折腾一大堆 Wine 的组件，有时还会面临封号的风险。反观现在拥抱全平台的吃相，让人忍俊不禁。
但真正改变的，其实很有限，因为爆火的原因不是厂商良心，拥抱开源，而是发现这个市场有的赚，本质还是逐利的，Agent 让厂商发现开放生态有的赚。工具功能很全，很方便，但我却更加担心了，工具会一直免费下去吗？我猜不会，等到官方的生态建立起来，我想就是回血的时候了。
值得高兴的是，这些工具还很烂，如果真的要用，还太早，比如 https://www.npmjs.com/package/@music163/ncm-cli
我无法相信官方的 TUI 居然吹能播放一半不到的歌，真是充满恶意的设计
-
LLM 改变了很多，比如 Agent 概念上半年爆火后，大厂们都争前恐后推出 CLI 产品，知乎，微信，网易云等等，这在几年前是完全无法想象的，那时想让 Linux 用上微信都很难，需要折腾一大堆 Wine 的组件，有时还会面临封号的风险。反观现在拥抱全平台的吃相，让人忍俊不禁。
但真正改变的，其实很有限，因为爆火的原因不是厂商良心，拥抱开源，而是发现这个市场有的赚，本质还是逐利的，Agent 让厂商发现开放生态有的赚。工具功能很全，很方便，但我却更加担心了，工具会一直免费下去吗？我猜不会，等到官方的生态建立起来，我想就是回血的时候了。
值得高兴的是，这些工具还很烂，如果真的要用，还太早，比如 https://www.npmjs.com/package/@music163/ncm-cli
我无法相信官方的 TUI 居然吹能播放一半不到的歌，真是充满恶意的设计
-
Was für eine Sch*** !
Ständig zerlegt sich die Whisper-Integration in KDEnlive. Kaum hat man die wieder geflickt, zerschießt das nächste Update sie wieder.
Mal eben Untertitel erzeugen, wird so zur Qual.
Ich prangere das an!
Danke für Ihre Aufmerksamkeit für dieses Thema!
-
Was für eine Sch*** !
Ständig zerlegt sich die Whisper-Integration in KDEnlive. Kaum hat man die wieder geflickt, zerschießt das nächste Update sie wieder.
Mal eben Untertitel erzeugen, wird so zur Qual.
Ich prangere das an!
Danke für Ihre Aufmerksamkeit für dieses Thema!
-
Was für eine Sch*** !
Ständig zerlegt sich die Whisper-Integration in KDEnlive. Kaum hat man die wieder geflickt, zerschießt das nächste Update sie wieder.
Mal eben Untertitel erzeugen, wird so zur Qual.
Ich prangere das an!
Danke für Ihre Aufmerksamkeit für dieses Thema!
-
I made this #YouTube video for work using #OpenSource. #Audacity with #OpenVINO to clean up the #Audio #Inkscape for the lower thirds and #Kdenlive to edit everything. Oh, and #Whisper to automatically create the #subtitles, which is did an amazing job, better than Audacity and OpenVINO did. Only would have been better if I did it on my home #kubuntu desktop but open source #software is the best and all ran on my Windows workstation.
-
La dettatura vocale arriva su Linux con un progetto open source che punta tutto su privacy, velocità e semplicità d'uso. Scopri come funziona Vocalinux e perché potrebbe diventare uno strumento indispensabile. #Linux #OpenSource #VoiceTyping #Privacy #Wayland #Whisper
-
Почему Cluely и другие плохо слышат русских айтишников: разбор того, как Whisper ломается и что мы сделали с этим
В январе я купил подписки на Cluely, Final Round AI и Sensei. Хотел посмотреть как они справляются с русским айти-собесами, раз уж все три заявляют о поддержке русского. Подключил по очереди к тестовому звонку в Телемосте (сомневаюсь, что платформа имела роль, но все же), прогнал одну и ту же запись: Senior Python backend разработчик, 45 минут, стек FastAPI + PostgreSQL + Kafka + Kubernetes. Обычный русский спикер, если важно - из Москвы, с речью проблем не было, нормальный микрофон Все три выдали транскрипт и все три провалились, как неожиданно.. "Кафка" в половине случаев становилась "как-то" или "кофта". "Кубернетис" превращался в "губер нет тест". "Сабскрайбер патерн" - в "саб скрайп патерн". "Middleware для CSRF" - "мидл-вер для си эс эр эф" - это еще норм Проблема не в том, что человек говорил по-русски, и не в том, что Whisper не умеет русский (сноска: хорошо не умеет). Whisper умеет русский нормально, около 9.8% WER на Common Voice. Проблема в другом: русскоязычный айтишник не говорит ни на чистом русском, ни на чистом английском. Он говорит на гибриде: русская грамматика плюс английские термины плюс своеобразное произношение этих терминов плюс местами свой жаргон вроде "гошечки" и "крудошлёпа" Этот гибрид ни один из популярных STT не держит. Потому что его в тренировочных данных почти нет Разбираю ниже, как устроена эта проблема, что с ней делают конкуренты (почти ничего), и что сделали мы
https://habr.com/ru/articles/1026778/
#Whisper #STT #speechtotext #finetuning #LoRA #ASR #NLP #распознавание_речи #русский_язык #codeswitching
-
🗣️🎤📝 :linux: Speech to Text and Text to Speech on GNU/Linux :disability_flag: 📝🔊💻
Why This Matters to Me (and Maybe You Too)
If you’re anything like me—a Linux user who counts on voice typing and TTS because of visual impairment—you know that accessibility is not a luxury, it’s a necessity. Speaking from experience as someone who depends on voice typing (and TTS) , the quest for a seamless, local, FLOSS speech-to-text (STT) setup on Linux can be frustrating.
Here’s how you can succeed with modern tools using Linux. FLOSS means freedom and privacy; working locally means real control.
Let’s dive in! I’ll tell you what I’ve learned and what I use—and hope you’ll share your favorite tools or tips!System-Wide Voice Keyboard: Speak Directly in Any App
Want to speak and have your words typed wherever your cursor is—be it a terminal, browser, chat, or IDE? Here’s what actually works and how it feels day-to-day:
- Speak to AI (Offline, Whisper-based, global hotkeys)
This tool is my current go-to. It uses Whisper locally, lets you use global hotkeys (configurable) to type into any focused window, and doesn’t need internet. Runs smoothly on X11 and Wayland; just takes a bit of setup (AppImage available!).
GitHub Repo https://github.com/AshBuk/speak-to-ai) | Dev.to Post https://dev.to/ashbuk/i-built-an-offline-voice-typing-app-for-linux-speak-to-ai-3ab5)- DIY: RealtimeSTT + PyAutoGUI
For the true tinkerers, RealtimeSTT plus a Python script lets you simulate keystrokes. You control every step, can lower latency with your tweaks, but you’ll need to be comfortable with scripting.
RealtimeSTT Guide https://github.com/KoljaB/RealtimeSTT#readme)- Handy (Free/Libre, offline, Whisper-based, acts as a keyboard)
I’ve read lots of positive feedback on Handy—even though I haven’t tried it myself. The workflow is simple: press a hotkey, speak, and Handy pastes your text in the active app. It’s fully offline, works on X11 and Wayland, and gets strong accuracy thanks to Whisper.
Heads up: Handy lets you pick your own shortcut key, but it actually overrides the keyboard shortcut for start/stop recording. That means it can clash with other tools that depend on major shortcut combos—including Orca’s custom keybindings if you use a screen reader. If your workflow relies on certain shortcuts, this might need adjustment or careful planning before you commit.
GitHub Repo https://github.com/cjpais/Handy) | Demo https://handy.computer)Real-Time Transcription in a Window (Copy/Paste Workflow)
If you’re okay with speaking into a dedicated app, then copying, these options offer great GUIs and power features:
- Speech Note by @mkiol https://mastodon.social/@mkiol
FLOSS, offline, multi-language GUI app—perfect for quick notes and batch transcription. Not a system-wide keyboard, but super easy to use and works on both desktops and Linux phones.
Flathub https://flathub.org/apps/net.mkiol.SpeechNote | LinuxPhoneApps https://linuxphoneapps.org/apps/net.mkiol.speechnote/)- WhisperLive (by Collabora)
Real-time transcription in a terminal or window—great for meetings, lectures, and captions. Manual copy/paste required to get the text to other apps.
GitHub Repo https://github.com/collabora/WhisperLive)More Tools for Tinkerers
If you like building your own or want extra control, check out:
- Vosk: Lightweight, lots of language support. GitHub https://alphacephei.com/vosk/)
- Kaldi: Powerful, best for custom setups. Website https://kaldi-asr.org/)
- Simon: Voice control automation. Website https://simon-listens.org/)
- voice2json: Phrase-level and command recognition. GitHub https://github.com/synesthesiam/voice2json)Pro Tips
- Desktop Environment: X11 vs. Wayland affects how keyboard hooks and app focus actually operate.
- Ready-Made vs. DIY: If you want plug-and-play, try Speech Note or Handy first. Into automation or customization? RealtimeSTT is perfect.
- Follow the Community: @thorstenvoice offers tons of open-source voice tech insights.Screen Reader Integration
Looking for robust screen reader support? Linux has you covered:
- Orca (GNOME/MATE): The most customizable GUI screen reader out there. The default voice (eSpeak) is robotic, but you can swap it for something better and fine-tune verbosity so it reads only what matters.
- Speakup: Console-based, ideal for terminal.
- Emacspeak: The solution for Emacs fans.💡 Orca is part of my daily toolkit. It took time to get the settings just right (especially verbosity!) but it’s absolutely worth it. If you use a screen reader—what setup makes it bearable or even enjoyable for you?
Final Thoughts
If you’re starting from scratch, try Handy for direct typing (just watch those shortcuts if you use a screen reader!) or Speech Note for GUI-based transcription. Both are privacy-friendly, local, and accessible—ideal for everyday Linux use.
Is there a FLOSS gem missing here?
Sharing what works (and what doesn’t!) helps the entire community.Resources:
Speech Note on Flathub https://flathub.org/apps/net.mkiol.SpeechNote
Handy GitHub https://github.com/cjpais/Handy
Speak to AI Guide https://dev.to/ashbuk/i-built-an-offline-voice-typing-app-for-linux-speak-to-ai-3ab5
RealtimeSTT https://github.com/KoljaB/RealtimeSTT#Linux #SpeechToText #FLOSS #Accessibility #VoiceKeyboard #ScreenReader #Whisper #Handy #SpeechNote #OpenSource #Community #voicetyping #LocalSTT #TTStools #SpeechRecognition #A11y #Linuxtools #Voicekeyboard #Whisper #Handy #speech-to-text #SpeechNote #review #ScreenReaders #ORCA #FOSS
-
📊 Some of the results from the Audacity OpenVINO Whisper Transcription study:
- 🕒 Average duration per model: Base (5.86s), Small (15.45s), Medium (39.03s), Large-v3 (81.33s).
- 🏅 Top scorer for accuracy: Small model!🔗 See the complete analysis: https://medium.com/@chribonn/comparing-audacitys-openvino-whisper-transcription-llms-c80c14834c1c
#DataScience #OpenVINO #Whisper #Audacity #TechInsights #TTMO -
HowTo Explains how to install and activate the OpenVINO plug-in in Audacity. This LLM is installed locally on your computer meaning that nothing gets transmitted to a remote servers. After installation, Audacity gains music separation, AI noise suppression, music generation and audio transcription. #Audacity #OpenVino #TTMO #HowTo #Intel #AI #Whisper, #Transcription, #SpeechToText, #MusicGeneration, #NoiseSuppression
https://www.alanbonnici.com/2025/04/howto-install-openvino-ai-plug-in-in.html
-
Have you ever wanted to use a #LLM as one step in a workflow?
We integrated #GPT into the open-source analysis platform #useGalaxy, where you can link GPT to several thousand other tools, add more attachments for analysis and make your research reproducible.
https://galaxyproject.org/news/2024-09-02-chat-gpt/
In our example, we uploaded an audio file and used #Whisper to convert it into text, cut out the moderation, and prompted chatGPT to translate it into German.
-
MacWhisper: Mac-App nimmt auf, transkribiert und macht Gesprochenes kopierbar
Die Anzahl der Anwendungen, die Projekte mit Bestandteilen Künstlicher Intelligenz auf den Mac bringen, nimmt seit wenigen Wochen deutlich zu. Auf die zahlreichen Applikationen, die zuerst den KI-Bildgeneratoren Stable Diffusion auf den Mac brachten, folgten erste Menüleisten-Apps, die den schlauen Chat-Bot ChatGPT in der Mac-Menüleiste ablegten. Die Macher des Übersetzungsdienstes DeepL lancierten zwischenzeitlich eine neue […]
-
#Matrix #Omega #Alex7000 #Xoanon #TheMagicMovieMachine #SCAPINA #Orac #Zen #TheOracle #Vanessa38–24–36 #C.O.R.A. #Mentalis #Dr.Theopolis #TheVortex #Gambit #Shyrka #Slave #CML #KITT #KARR #R.A.L.F. #TeletraanI #VectorSigma #BriantheBrain #Compucore #SID #Synergy #Box #LCARS #Albert #Crossover #MagicVoice #OMNSS #Priscilla #Holly #Gordon8000 #Queeg #Hilly #TalkieToaster #TheRevolvingToilet #Sandy #TheUltimaMachine #Ziggy #P.J. #HARDAC #COS #CAS #NICOLE #SELMA #CentSys #MetroNet #H.E.L.E.N. #SharonApple #TheMagiMelchiorBalthasarCaspar #Eve #L.U.C.I-U.N.I.C.E #EmergencyMedicalHologram #Alice #Long-termMedicalHolographicprogram, #VicFontaine, #GilliamII #QuadraplexT-3000Computer #Memorymatic #MPU #Starfighter31 #Computer #P.A.T. #D.E.C.A. #BlackBetty #KarenPlankton #Andromeda #Comp-U-Comp #Caravaggio #C.H.A.D. #GLADIS #CybergirlXandaIsaac #Computer #SAINT #AuraMorganna #Vox #AI #Wirbelwind #Delphi #Sheila/F.I.L.S.S. #OoGhiJMIQtxxXA #XANA #Survive #C.A.R.R. #D.A.V.E. #TheOmnitrix #Solty/Dike #Eunomia #Eirene #Bournemouth #S.O.P.H.I.E. #Scylla #FETCH!3000 #S.A.R.A.H. #TheIntersect #MrSmith #Pear #TheTurk #KITT #POD #ISIS #VenjixVirus #VYVAIVAI #Whisper #Aya #TheMachine #R.A.C.I.S.T. #Samaritan