Need help with real-time sentence segmentation in faster-whisper for simultaneous translation
#6
by
ekremabiii
- opened
I'm working on a real-time translation project where I'm using faster-whisper for speech-to-text conversion. The issue I'm facing is that faster-whisper's segment-based transcription sometimes sends multiple sentences to the translator simultaneously when silence is detected. I need a way to process each sentence individually in real-time as they are spoken, rather than receiving multiple sentences bundled together. Is there a way to achieve more granular sentence-by-sentence processing with faster-whisper, or should I implement a custom solution for sentence segmentation?