-
Notifications
You must be signed in to change notification settings - Fork 6
/
api.py
200 lines (164 loc) · 6.65 KB
/
api.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
import base64
import os
import torch
import time
# vits
import vits_commons
import vits_utils
from vits_models import SynthesizerTrn
# so-vits
from inference.infer_tool import Svc
from inference import infer_tool
from inference import slicer
import numpy as np
import io
import soundfile
import logging
# common
from text.symbols import symbols
from text import text_to_sequence
import json
# flask
from flask_cors import CORS
from flask import Flask, jsonify, request
logging.getLogger().setLevel(logging.ERROR)
# load config.json
config = None
with open("config.json", "r") as f:
config = json.load(f)
speaker_VITS = 22 # 22 and 99 was good
sovits_models = config["sovits_models"]
# check if os is windows
if os.name == 'nt':
os.environ["PHONEMIZER_ESPEAK_PATH"] = config["phonemizer"]["PHONEMIZER_ESPEAK_PATH"]
os.environ["PHONEMIZER_ESPEAK_LIBRARY"] = config["phonemizer"]["PHONEMIZER_ESPEAK_LIBRARY"]
# defining flask app
app = Flask("ganyuTTS")
CORS(app)
def get_text(text, hps):
text_norm = text_to_sequence(text, hps.data.text_cleaners)
if hps.data.add_blank:
text_norm = vits_commons.intersperse(text_norm, 0)
text_norm = torch.LongTensor(text_norm)
return text_norm
def initModels():
# loading VITS model
global hps
hps = vits_utils.get_hparams_from_file("./configs/ljs_base.json")
global net_g
net_g = SynthesizerTrn(
len(symbols),
hps.data.filter_length // 2 + 1,
hps.train.segment_size // hps.data.hop_length,
**hps.model)
_ = net_g.eval()
global hps_ms
hps_ms = vits_utils.get_hparams_from_file("./configs/vctk_base.json")
global net_g_ms
net_g_ms = SynthesizerTrn(
len(symbols),
hps_ms.data.filter_length // 2 + 1,
hps_ms.train.segment_size // hps.data.hop_length,
n_speakers=hps_ms.data.n_speakers,
**hps_ms.model)
_ = net_g.eval()
_ = vits_utils.load_checkpoint("models/pretrained_vctk.pth", net_g_ms, None)
# send model to gpu
if torch.cuda.is_available():
net_g.cuda()
net_g_ms.cuda()
# Loading SO-VITS models
global sovits_models
for model in sovits_models:
sovits_models[model]["model"] = Svc(sovits_models[model]["model_path"], sovits_models[model]["config_path"])
#if the value's are set already, use them instead of the parameters
# check if sovits_models[model]["trans"] exists
if "trans" not in sovits_models[model].keys():
sovits_models[model]["trans"] = 0
if "auto_predict_f0" not in sovits_models[model].keys():
sovits_models[model]["auto_predict_f0"] = False
if "cluster_infer_ratio" not in sovits_models[model].keys():
sovits_models[model]["cluster_infer_ratio"] = 0
if "noice_scale" not in sovits_models[model].keys():
sovits_models[model]["noice_scale"] = 0.4
if "pad_seconds" not in sovits_models[model].keys():
sovits_models[model]["pad_seconds"] = 0.5
def generate_VITS(text, sid=22):
if (sid == 22):
length_scale = 1.2
else:
length_scale = 1.1
stn_tst = get_text(text, hps_ms)
print("VITS start on voice " + str(sid))
with torch.no_grad():
x_tst = stn_tst.unsqueeze(0)
x_tst_lengths = torch.LongTensor([stn_tst.size(0)])
sid = torch.LongTensor([speaker_VITS])
audio = net_g_ms.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=.667, noise_scale_w=0.8, length_scale=length_scale)[0][0,0].data.float().cpu().numpy()
print("VITS done")
# clean cuda memory cache
torch.cuda.empty_cache()
return audio, hps_ms.data.sampling_rate
def generate_SO_VITS(audio, sr, modelname="ganyu"):
global sovits_models
chunks = slicer.cut2(audio, sr=sr)
audio_data, audio_sr = slicer.chunks2audio2(audio=audio, sr=sr, chunks=chunks)
audio = []
for (slice_tag, data) in audio_data:
print(f'#=====segment start, {round(len(data) / audio_sr, 3)}s======')
length = int(np.ceil(len(data) / audio_sr * sovits_models[modelname]["model"].target_sample))
if slice_tag:
print('jump empty segment')
_audio = np.zeros(length)
else:
# padd
pad_len = int(audio_sr * sovits_models[modelname]["pad_seconds"])
data = np.concatenate([np.zeros([pad_len]), data, np.zeros([pad_len])])
raw_path = io.BytesIO()
soundfile.write(raw_path, data, audio_sr, format="wav")
raw_path.seek(0)
out_audio, out_sr = sovits_models[modelname]["model"].infer(modelname, sovits_models[modelname]["trans"], raw_path,
cluster_infer_ratio=sovits_models[modelname]["cluster_infer_ratio"],
auto_predict_f0=sovits_models[modelname]["auto_predict_f0"],
noice_scale=sovits_models[modelname]["noice_scale"]
)
_audio = out_audio.cpu().numpy()
pad_len = int(sovits_models[modelname]["model"].target_sample * sovits_models[modelname]["pad_seconds"])
_audio = _audio[pad_len:-pad_len]
audio.extend(list(infer_tool.pad_array(_audio, length)))
# clean cuda memory cache
torch.cuda.empty_cache()
return audio, sovits_models[modelname]["model"].target_sample
@app.route('/tts', methods=['POST'])
def tts():
start = time.time()
text = request.form.get('text')
# get sid too, but only if it's provided
if 'sid' in request.form:
sid = int(request.form.get('sid'))
else:
sid = 22
if 'sid2' in request.form:
sid2 = str(request.form.get('sid2'))
print("Got text from client: " + text)
audio, sr = generate_VITS(text, sid)
audio, sr = generate_SO_VITS(audio, sr, sid2)
file_object = io.BytesIO()
soundfile.write(file_object, audio, sr, format="wav")
file_object.seek(0)
file_string = file_object.read()
audio = base64.b64encode(file_string).decode('utf-8')
print("Done in " + str(time.time() - start) + " seconds")
return jsonify({'audio': audio})
def main():
initModels()
# warmup
print("Warming up...")
audio, sr = generate_VITS("Hey Commander! Universal Cartographics service has been paused as you ordered!", 22)
# save audio into temp as api_warmup_debug.wav
soundfile.write("./tmp/api_warmup_debug.wav", audio, sr, format="wav")
audio, sr = generate_SO_VITS(audio, sr, "ganyu")
soundfile.write("./tmp/api_warmup_debug2.wav", audio, sr, format="wav")
if __name__ == "__main__":
main()
app.run(host='0.0.0.0', port=4111, debug=False)