transcribe
Transcribe audio to text using AI speech-to-text models.
Supports OpenAI (whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe),
ElevenLabs (scribe_v1), Deepgram (nova-3), Groq (whisper-large-v3,
whisper-large-v3-turbo, distil-whisper-large-v3-en), and Cartesia
(ink-whisper). Outputs JSON with text, timestamps, language, and duration.
Usage
1egaki transcribe [audio]
Arguments
| Argument | Required | Description |
[audio] | No | audio |
Options
| Option | Default | Description |
-m, --model [model] | - | Transcription model ID. If omitted, shows an interactive picker (or uses default in non-TTY mode) |
-o, --output [path] | - | Output file path for the JSON result |
--language [lang] | - | ISO 639-1 language hint (e.g. en, es, fr). Helps accuracy for known languages |
--stdin | - | Read audio from stdin instead of a file path |
--stdout | - | Write only the plain text transcript to stdout (no JSON metadata) |
Global Options
| Option | Default | Description |
-h, --help | - | Display this message |
-v, --version | - | Display version number |
Examples
1# Transcribe an audio file with default model
1egaki transcribe recording.mp3
1# Use a specific model
1egaki transcribe interview.wav -m gpt-4o-transcribe
1# Groq for fast, cheap transcription
1egaki transcribe podcast.mp3 -m whisper-large-v3
1# Get plain text only
1egaki transcribe recording.mp3 --stdout
1# Save JSON result to file
1egaki transcribe recording.mp3 -o transcript.json
1# Pipe audio from another command
1ffmpeg -i video.mp4 -f mp3 - | egaki transcribe --stdin -m whisper-1