Setup & Installation
openclaw skills install @sfkiwi/local-vosk
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
npx clawhub install local-vosk
Version History
v1.0.1Feb 10, 2026 - Version note by skill developer:
Added Telegram voice message use case + improved docs
What This Skill Does
Runs speech-to-text locally using the Vosk toolkit. Works fully offline after a one-time model download and supports any audio format ffmpeg can decode. No API keys, accounts, or cloud services needed.
Unlike cloud STT services, it processes audio locally with no usage costs, no data sent to third parties, and no API rate limits.
When to use it
- Transcribing Telegram .ogg voice messages
- Converting recorded meeting audio to text
- Batch transcribing audio files without internet access
- Processing mp3, wav, m4a, and webm formats locally
- Offline transcription on restricted or air-gapped machines
Example Workflow
Here's how your AI assistant might use this skill in practice.
INPUT
User asks: transcribe this Telegram voice message (voice_message.ogg)
AGENT
- 1Receives the .ogg audio file path
- 2Converts audio to 16kHz mono WAV using ffmpeg
- 3Loads the local Vosk model from ~/vosk-models
- 4Processes the audio stream and produces a transcript
OUTPUT
Plain text transcript of the voice message
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.
vosk Python packageVosk language model downloaded from alphacephei.com/vosk/models