Vocoder - Githubissues

agostini01 / Real-Time-Voice-Cloning

Clone a voice in 5 seconds to generate arbitrary speech in real-time

0 stars 4 forks source link

Vocoder #7

Open agostini01 opened 4 years ago

agostini01 commented 4 years ago

[ ] Profiling
[ ] Pruning
[x] Quantization

PragyanJain commented 4 years ago

Pytorch quantization works only on CPU not supported on GPU.

Quantization Results- Size of the model before quantization(MB) : 17.972838 Size of the model after quantization(MB) : 15.343152

Inference is based on one sample input mel spectrogram: Time elapsed with pertained model(seconds) : 38.94 (Generation Rate 3.2 kHz) Time elapsed with quantized model(seconds) : 32.85 (Generation Rate 3.8 kHz)