tuna2134/sbv2-v2
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Merge pull request #15 from fumiama/master

Browse Source 
feat(webui): add mps support
This commit is contained in:
Stardust·减
2023-09-03 17:14:34 +08:00
committed by GitHub
parent ca4b39cdcd 5a6f824537
commit 3c93ea55fa
1 changed files with 24 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
webui.py
View File

@@ -1,3 +1,8 @@
import sys, os
if sys.platform == "darwin":
os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"
import torch import torch
import argparse import argparse
import commons import commons
@@ -9,6 +14,10 @@ from text.cleaner import clean_text
import gradio as gr import gradio as gr
import webbrowser import webbrowser
net_g = None
def get_text(text, language_str, hps): def get_text(text, language_str, hps):
norm_text, phone, tone, word2ph = clean_text(text, language_str) norm_text, phone, tone, word2ph = clean_text(text, language_str)
phone, tone, language = cleaned_text_to_sequence(phone, tone, language_str) phone, tone, language = cleaned_text_to_sequence(phone, tone, language_str)
@@ -21,6 +30,7 @@ def get_text(text, language_str, hps):
word2ph[i] = word2ph[i] * 2 word2ph[i] = word2ph[i] * 2
word2ph[0] += 1 word2ph[0] += 1
bert = get_bert(norm_text, word2ph, language_str) bert = get_bert(norm_text, word2ph, language_str)
del word2ph
assert bert.shape[-1] == len(phone) assert bert.shape[-1] == len(phone)
@@ -31,16 +41,19 @@ def get_text(text, language_str, hps):
return bert, phone, tone, language return bert, phone, tone, language
def infer(text, sdp_ratio, noise_scale, noise_scale_w, length_scale, sid): def infer(text, sdp_ratio, noise_scale, noise_scale_w, length_scale, sid):
bert, phones, tones, lang_ids = get_text(text, "ZH", hps,) global net_g
bert, phones, tones, lang_ids = get_text(text, "ZH", hps)
with torch.no_grad(): with torch.no_grad():
x_tst=phones.to(device).unsqueeze(0) x_tst=phones.to(device).unsqueeze(0)
tones=tones.to(device).unsqueeze(0) tones=tones.to(device).unsqueeze(0)
lang_ids=lang_ids.to(device).unsqueeze(0) lang_ids=lang_ids.to(device).unsqueeze(0)
bert = bert.to(device).unsqueeze(0) bert = bert.to(device).unsqueeze(0)
x_tst_lengths = torch.LongTensor([phones.size(0)]).to(device) x_tst_lengths = torch.LongTensor([phones.size(0)]).to(device)
del phones
speakers = torch.LongTensor([hps.data.spk2id[sid]]).to(device) speakers = torch.LongTensor([hps.data.spk2id[sid]]).to(device)
audio = net_g.infer(x_tst, x_tst_lengths, speakers, tones, lang_ids,bert, sdp_ratio=sdp_ratio audio = net_g.infer(x_tst, x_tst_lengths, speakers, tones, lang_ids, bert, sdp_ratio=sdp_ratio
, noise_scale=noise_scale, noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0,0].data.cpu().float().numpy() , noise_scale=noise_scale, noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0,0].data.cpu().float().numpy()
del x_tst, tones, lang_ids, bert, x_tst_lengths, speakers
return audio return audio
def tts_fn(text, speaker, sdp_ratio, noise_scale, noise_scale_w, length_scale): def tts_fn(text, speaker, sdp_ratio, noise_scale, noise_scale_w, length_scale):
@@ -58,7 +71,15 @@ if __name__ == "__main__":
args = parser.parse_args() args = parser.parse_args()
hps = utils.get_hparams_from_file(args.config) hps = utils.get_hparams_from_file(args.config)
device = "cuda:0" if torch.cuda.is_available() else "cpu" device = (
"cuda:0"
if torch.cuda.is_available()
else (
"mps"
if sys.platform == "darwin" and torch.backends.mps.is_available()
else "cpu"
)
)
net_g = SynthesizerTrn( net_g = SynthesizerTrn(
len(symbols), len(symbols),
hps.data.filter_length // 2 + 1, hps.data.filter_length // 2 + 1,