Real-Time Streaming Mel Vocoding with Low Latency Flow Matching

MelFlow is a streaming-capable Mel vocoder that runs at 16 kHz with an algorithmic latency of 32 ms and total latency of only 48 ms, demonstrated on a consumer-grade laptop GPU. getnews.me/real-time-streaming-mel-... #melflow #vocoding #tts

.@nvidia promoted #fugatto (foundational generative audio transformer opus),a #new #generative #ai #model which can #create any combination of #music, #voices & #sounds.a transformer model that builds on their prior work on #speech #modeling,#audio #vocoding & #understanding
youtu.be/qj1Sp8He6e4?...