home.social

#speech_to_text — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #speech_to_text, aggregated by home.social.

  1. I’m currently tinkering with an idea:
    An offline dictation app for Linux that transcribes spoken language using Whisper and optionally refines it using a locally running model (“Enhance” step: smooth out grammar, remove filler words). 😁

    As is my usual practice, I’m building it ~99% with LLM support.

    Since I’m not a programmer, but rather a proof-of-concept tinker. 😉

    For the first prototype:
    Hotkey → Record → Transcription → Text lands in one of 9 fixed output fields on the desktop. Cursor-precise insertion and a meetings mode will come later (if at all).

    Question for the group:
    Who on Linux would even be interested in something like this, or is there already something similar out there that I’ve overlooked? 🤔

    #Linux #Dictation #Speech_to_Text #Indie_Web #Idea #LLM #AI #App #Vibe_Coding #Transcription #Programming #Tinkering

  2. Murena /e/OS - Voice to Text

    Eine lokale Speech-to-Text-Lösung für Smartphones mit dem /e/OS-Custom ROM von Murena. Kann das funktionieren und wie gut ist die Qualität? Lest es hier.

    #Murena #Speech_to_Text #LLM #eOS #Linux

    gnulinux.ch/murena-e-os-voice-

  3. Texte flüstern mit Speed of Sound

    Mit "Speed of Sound" gibt es eine neue Anwendung, die Gesprochenes in geschriebenen Text umwandelt. Dabei kommt das kleine Whisper-LLM von OpenAI zum Einsatz. Wie gut das funktioniert, lest ihr hier.

    #STT #Whisper #OpenAI #Speech_to_Text #Linux

    gnulinux.ch/texte-fluestern-mi