Stopping Whisper’s Infinite Loops with Voice Activity Detection

AI transcription models like Whisper often hallucinate repetitive phrases during silence or background noise, leading to unusable outputs. Implementing Voice Activity Detection (VAD) or using domain-specific models like Sarvam AI is crucial for maintaining transcription accuracy in noisy, real-world audio.