nvidia

/

nemotron-speech-streaming-en-0.6b

like 526

Follow

NVIDIA 55.3k

Hey Team,
New to this model/architecture and still exploring around. But is it possible to have a TDT based model instead of RNN-T. From what I know, we get even faster inference and comparable (or better) accuracy with TDT compared to RNN-T.

Thanks in advance