elevenlabs-transcribe
by paulasjes
Speech & TranscriptionBrief
Transcribe audio to text using ElevenLabs.
Description
What This Skill Does
Transcribes audio and video files to text using ElevenLabs Scribe. Supports batch processing of local files, realtime streaming from URLs or microphones, and speaker diarization across 90+ languages.
Handles both batch and realtime streaming in one tool, with speaker diarization and word-level timestamps that most transcription wrappers omit.
When to Use It
- Transcribing recorded meetings with speaker labels
- Capturing live voice input during an agent session
- Converting podcast or radio streams to text in realtime
- Extracting subtitles from video files with word timestamps
- Batch transcribing interview recordings for analysis
Example Workflow
User asks: transcribe this meeting recording and identify who said what
- 1
Receives meeting.mp3 as input
- 2
Runs transcribe.sh meeting.mp3 --diarize --json
- 3
ElevenLabs Scribe API processes the audio and returns speaker-labeled segments
- 4
Parses JSON output to extract text grouped by speaker_id
- 5
Returns formatted transcript with speaker attribution
Full transcript with each speaker's lines labeled as speaker_0, speaker_1, etc., including word-level timestamps
Requirements
- ELEVENLABS_API_KEY environment variable
- ffmpeg installed on the system
Installation
clawhub install paulasjes/elevenlabs-transcribeRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.