|
@ -116,9 +116,7 @@ |
|
|
"source": [ |
|
|
"source": [ |
|
|
"waveglow_path = 'waveglow_old.pt'\n", |
|
|
"waveglow_path = 'waveglow_old.pt'\n", |
|
|
"waveglow = torch.load(waveglow_path)['model']\n", |
|
|
"waveglow = torch.load(waveglow_path)['model']\n", |
|
|
"waveglow.cuda().half()\n", |
|
|
|
|
|
"for k in waveglow.convinv:\n", |
|
|
|
|
|
" k.float()" |
|
|
|
|
|
|
|
|
"waveglow.cuda()" |
|
|
] |
|
|
] |
|
|
}, |
|
|
}, |
|
|
{ |
|
|
{ |
|
@ -205,7 +203,7 @@ |
|
|
], |
|
|
], |
|
|
"source": [ |
|
|
"source": [ |
|
|
"with torch.no_grad():\n", |
|
|
"with torch.no_grad():\n", |
|
|
" audio = waveglow.infer(mel_outputs_postnet.half(), sigma=0.666)\n", |
|
|
|
|
|
|
|
|
" audio = waveglow.infer(mel_outputs_postnet, sigma=0.666)\n", |
|
|
"ipd.Audio(audio[0].data.cpu().numpy(), rate=hparams.sampling_rate)" |
|
|
"ipd.Audio(audio[0].data.cpu().numpy(), rate=hparams.sampling_rate)" |
|
|
] |
|
|
] |
|
|
} |
|
|
} |
|
|