aux4/transcribe

Transcribe audio and video files to text with timestamps. Accepts local files or URLs and outputs a JSON array of timestamped text segments.

Installation

aux4 aux4 pkger install aux4/transcribe

Quick Start

aux4 transcribe recording.mp3

aux4 transcribe https://example.com/audio.wav

Configuration

The package uses the OpenAI Whisper API for transcription. Provide credentials via environment variable or a config file:

Environment Variable

export OPENAI_API_KEY=sk-...
aux4 transcribe recording.mp3

Config File

config:
  transcribe:
    model:
      type: openai
      api: default

aux4 transcribe recording.mp3 --configFile config.yaml --config transcribe

The api field supports:

• default — Uses the standard OpenAI API (OPENAI_API_KEY)
• codex — Uses the Codex API endpoint (CODEX_API_KEY)

Output Format

The output is a JSON array of timestamped segments:

[
  {
    "time": "0:00",
    "seconds": 0,
    "text": "Welcome to the assembly guide."
  },
  {
    "time": "0:15",
    "seconds": 15,
    "text": "Start by laying out all the parts."
  }
]

Supported Formats

Any audio or video format supported by the OpenAI Whisper API:

• Audio: mp3, mp4, mpeg, mpga, m4a, wav, webm
• Video: mp4, webm (audio track is extracted)

Commands

aux4 transcribe <file>

Transcribe an audio or video file to text with timestamps.

| Flag | Description | Default | |------|-------------|---------| | file | Path or URL to the audio/video file | (required) | | --configFile | Configuration file path | | | --config | Configuration profile name | |