NVIDIA / tacotron2

Tacotron 2 - PyTorch implementation with faster-than-realtime inference
BSD 3-Clause "New" or "Revised" License
4.97k stars 1.37k forks source link

Input & output streaming #627

Open rumbleFTW opened 2 months ago

rumbleFTW commented 2 months ago

Is it possible to make this fully stream for both inputs and outputs such it can start generating once it sees the first word from the input stream. This will greatly decrease the TTFB and will unlock infinite possibilities for real time applications.