Latecript

Latecript is a Python application that transcribes and translates audio from your input devices in real-time using the Speechmatics API. The application provides a Textual-based TUI for an interactive experience.

(Developped quick and dirty with the help of Copilot)

Features

• Real-Time Transcription: Capture audio and transcribe speech in real-time.
• Translation: Translate transcribed text into different languages.
• TUI Interface: Interactive Textual UI for settings and displaying logs.
• Configurable Settings: Read credentials and configuration from a local JSON file.

Requirements

• Python 3.12 or above
• UV (for dependency management)

Getting your audio output as a usable input.

MacOs

You can use BlackHole for audio loopback. It can be installed via brew:

brew install blackhole-2ch

While using the app, chose "BlackHole 2ch" as your sound output. In order to still listen to what your mac sound output you can define a multi-output device (with you favourite output device + BlackHole 2ch) in the Audio MIDI setup.

Speechmatics API key

You can generate your speechmatics API key from your user account in speechmatics.

Usage

To run Latecript, execute:

uv run latecript 

You can provide an alternative settings file via the command line:

uv run latecript --settings_file /path/to/your/settings.json

The settings file is a json file with the following structure:

{
  "speechmatics_api_key": "Your Speechmatics API Key",
  "output_device": "Blackhole 2ch",
  "transcription_language": "fr",
  "translation_language": "en"
}

License

This project is licensed under the MIT License. See LICENSE for details.

Contributing

Contributions are welcome! Please open issues and submit pull requests for improvements and bug fixes.

This app is not meant to be maintained and was developped in a quick and dirty way.

Acknowledgements

• Speechmatics for the API and SDK.
• Textual for the TUI framework.
• BlackHole for audio loopback driver for mac.