/live-translation

Translate the audio input using readily available machine learning tools and text-to-speech technology.

Primary LanguagePythonMIT LicenseMIT

live-translation

Translate the audio input using readily available machine learning tools and text-to-speech technology. This project is divided into parts, one stand alone Python script that collects the audio, translate and transfer the text to the web application.

  graph TD 
    A[Speaker] --> B[Auto Translator service] 
    B --> C[Web Application] 
    C --> D[Mobile devices]
Loading

How does it work?

Initialisation

The normal flow involves someone starting an live event and consequently generating a QR Code that points to the event.

Access

Spectators interested in the translation of the event can scan the QR code and access the live translation and select the target language.

Broadcast

The Translator service will receive chunks of audio, translate then and transmit them to the Web Application and this one will broadcast to the connected expectators.

sequenceDiagram
    Administrator->>+WebApp: 1. Start event.
    WebApp->>WebApp: 2. Generate QR code.
    WebApp->>-Administrator: 3. QR code.
    Spectator->>+Spectator: 4. Scan QR code.
    Spectator->>WebApp: 5. Access live translation (websocket).
    Spectator->>WebApp: 5.2. Select language.
    Presenter->>+AutoTranslator: 6. Live audio.
    AutoTranslator->>+WebApp: 7. Translation.
    WebApp->>-Spectator: 8. Translation.
Loading