Skip to content

Latest commit

 

History

History
76 lines (50 loc) · 3.17 KB

readme.md

File metadata and controls

76 lines (50 loc) · 3.17 KB
Raw

Real-Time Speech Translation with Virtual Microphone

This project provides a real-time speech translation system that captures audio from your microphone, transcribes it to text, translates the text into another language, and outputs the translated speech through a virtual microphone. The virtual microphone (e.g., BlackHole or VB-Audio Cable) can be used in applications like Google Meet, Zoom, or Microsoft Teams for live translated speech.

Requirements

  • Node.js
  • Deepgram API (for speech-to-text transcription)
  • Google Cloud Translation API (for translating text)
  • Virtual Audio Cable or BlackHole (for routing translated audio into other apps)

How It Works

  1. Audio Capture:

    • The app captures audio from the user's microphone using the Web Audio API.
    • This audio is processed and streamed to Deepgram for real-time transcription.

  2. Transcription:

    • Transcriptions are received via a WebSocket from Deepgram's API.
    • The transcribed text is displayed and processed for translation.

  3. Translation:

    • Transcriptions are sent to the Google Cloud Translation API for real-time translation into a specified target language.

  4. Text-to-Speech (TTS):

    • The translated text is converted to speech using Deepgram's TTS WebSocket service.

  5. Virtual Microphone Output:

    • The translated audio is routed to a virtual microphone (e.g., BlackHole or VB-Audio Cable), which can then be used as the input in video conferencing applications.

Data Flow (Piping):

  1. Input from Microphone:

    • Audio is captured from the user’s microphone (excluding virtual mics to prevent feedback loops).

  2. Transcription and Translation:

    • Audio is sent to the Deepgram API for transcription.
    • The transcribed text is sent to Google’s Translation API for translation.

  3. Queueing and Playback:

    • Translated text is sent to the Deepgram TTS service for speech synthesis.
    • The synthesized speech is queued for playback.

  4. Virtual Audio Output:

    • The output is routed to the virtual microphone, making it available for apps like Zoom, Teams, or Google Meet.

Virtual Microphone Setup

To use the translated speech in video conferencing apps, you'll need to set up a virtual audio device:

Once the virtual microphone is installed:

  1. Select the virtual microphone as the input in your video conferencing app to stream the translated speech live.

How to Run

  1. Clone the repository and install dependencies:

    npm install

  2. Environment variables are already set up with testing values for demonstration purposes. These include temporary keys for Deepgram and Google Cloud APIs.

  3. Start the app:

    npm run start

  4. Open the app in a browser, select the language, and start transcription and translation using the UI.

Demo video (this is what the other person in the meeting will hear):

realtime-translation-demo.mp4