salute-speech
by chorus12
Media & StreamingBrief
Transcribe audio files using Sber Salute Speech async API.
Description
What This Skill Does
Transcribes audio and video files to timestamped text using Sber's Salute Speech async REST API. Supports Russian, English, Kazakh, Kyrgyz, and Uzbek. Produces a formatted transcript file and a raw JSON file with full timing and confidence data.
Covers Central Asian languages (Kazakh, Kyrgyz, Uzbek) that most mainstream STT services don't support.
When to Use It
- Transcribing Russian-language meeting recordings
- Converting interview audio to searchable text
- Processing call center recordings with a telephony-optimized model
- Transcribing Kazakh or Uzbek audio files
- Extracting timestamped captions from podcast episodes
Example Workflow
User asks: Transcribe this Russian meeting recording: meeting.mp3
- 1
Reads SALUTE_AUTH_DATA from the host environment
- 2
Fetches a short-lived auth token from the Salute Speech API
- 3
Uploads meeting.mp3 and submits an async transcription job
- 4
Polls for the result until the job completes
- 5
Saves the timestamped transcript and raw JSON to the output directory
A formatted .txt file with timestamped segments (e.g. [00:01 - 00:20]: ...) and a raw JSON file with hypothesis and confidence data
Requirements
- SALUTE_AUTH_DATA environment variable — Base64-encoded client_id:client_secret from https://developers.sber.ru/studio/
Installation
clawhub install chorus12/salute-speechRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.