Speech Translation Demo

Speak into the microphone and your speech will be transcribed and translated segment-by-segment. (Recording is limited to 15 seconds.)

Note: The translation and speech synthesis occur automatically after recording.

Speech Translation Demo with Automatic TTS and Restart Option

This demo performs the following:

1. Accepts up to 15 seconds of audio recording from the microphone.
2. Uses OpenAI’s Whisper model to transcribe the speech.
3. Splits the transcription into segments and translates each segment on-the-fly using Facebook’s M2M100 model.
4. Streams the cumulative translation output to the user.
5. Automatically converts the final translated text to speech using gTTS.
6. Provides a "Restart Recording" button (located just below the recording section) to reset the audio input, translated text, and TTS output.

Note: True real-time translation (i.e. while speaking) requires a continuous streaming solution which is not provided by the standard browser microphone input.