Voice Transcription for Emacs using OpenAI’s Realtime API

Requirements

• Emacs 27.1 or later
• ffmpeg
• PulseAudio (currently supports only Linux systems with PulseAudio)
• OpenAI API Key

Installation

MELPA

Coming soon.

Manual

Clone this repository and add its path to `load-path`:

“`elisp (add-to-list ‘load-path “/path/to/ok-audio-transcription”) (require ‘ok-audio-transcription) “`

Configuration

Set your OpenAI API token:

“`elisp (setq ok-audio-transcription–openai-token “your-api-token”) “`

Usage

`M-x ok-audio-transcription–record-dwim`

Start/stop recording (press once to start, again to stop)

`M-x ok-audio-transcription–transcribe-file`

Transcribe an existing audio file

The transcription result will be inserted at point.

Define a shortcut:

“`elisp (define-key global-map (kbd “C-x R”) ‘ok-audio-transcription–record-dwim) “`

Limitations

• Currently supports only Linux systems with PulseAudio
• Maximum recording length defaults to 180 seconds (modifiable via `ok-audio-transcription–max-recording-seconds`)
• Requires an active internet connection
• Utilizes OpenAI’s API (incurs costs)

License

AGPL-3.0. See LICENSE.

Credits

Inspired by Simon Safar’s blog post.