Rayhane-mamah / Tacotron-2

DeepMind's Tacotron-2 Tensorflow implementation
MIT License
2.27k stars 905 forks source link

output_per_step around 2~5,but final-mel-output is only one frame #380

Open Liujingxiu23 opened 5 years ago

Liujingxiu23 commented 5 years ago

why is output_per_step around 2~5, but the decoder output is reshaped to only one frame? Does this slow the synthesize speed? Does anyone test the frame of final output == output_per_step?

xxoospring commented 5 years ago

一代我提交了问题,麻烦你看看,谢谢