fix: use librosa instead of torchaudio for VAD (torchcodec not installed), fix audios deprecation

add fast captioning module (CLAP + faster-whisper + Silero VAD), update deps