transcribe-assemblyai

name	transcribe-assemblyai
description	Transcribe audio via AssemblyAI with word-level timestamps and speaker diarization. Best for meetings, interviews, and anything needing speaker labels or timecodes. Use when the user asks for a timestamped transcript, diarized transcript, or multi-speaker transcription.

Transcribe (AssemblyAI, timestamped + diarized)

Produce a transcript with word-level timestamps and speaker turns.

Requirements

• ASSEMBLYAI_API_KEY env var (name configurable in config.json).
• Source audio accessible — AssemblyAI accepts URL or uploaded file.

How

1. Upload file: POST https://api.assemblyai.com/v2/upload (bytes of audio, Authorization: <key>).
2. Create transcript: POST https://api.assemblyai.com/v2/transcript with

   {
     "audio_url": "<upload_url>",
     "speaker_labels": true,
     "punctuate": true,
     "format_text": true
   }
   

3. Poll GET /v2/transcript/<id> until status == "completed".
4. Render output.

Speaker name mapping

If <source_stem>_speakers.json exists (from extract-speaker-samples), remap AssemblyAI's A/B/C labels to user-provided names.

Output convention

Save to <source_stem>.timestamped.md. Format:

[00:00:03] **Alice:** Welcome to the call.
[00:00:07] **Bob:** Thanks for having me.

Also save raw AssemblyAI JSON to <source_stem>.assemblyai.json for downstream use.

Cost

Roughly $0.37/hr with speaker labels. Flag this if the audio is very long (>4 hours).