-
Notifications
You must be signed in to change notification settings - Fork 14
/
Copy pathgenerate_tiktok.py
169 lines (143 loc) · 6.46 KB
/
generate_tiktok.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
import os
import numpy as np
import torch as t
import torch.nn as nn
import torch.nn.functional as F
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter
import yaml
import time
import argparse
from pathlib import Path
import jukebox.utils.dist_adapter as dist
from jukebox.hparams import Hyperparams
from jukebox.utils.torch_utils import empty_cache
from jukebox.utils.audio_utils import save_wav, load_audio
from jukebox.make_models import make_vae_model
from jukebox.utils.sample_utils import split_batch, get_starts
from jukebox.utils.dist_utils import print_once
import fire
import librosa
import soundfile as sf
import noisereduce as nr
from d2m.dataset import TiktokDataset
from d2m.d2m_modules_tiktok import vqEncoder_high, vqEncoder_low, Discriminator, motion_encoder, Audio2Mel
from d2m.utils import save_sample
def parse_args():
parser = argparse.ArgumentParser()
#parser.add_argument("--save_path", required=True)
parser.add_argument("--load_path", default='./logs')
parser.add_argument("--model", default='5b')
parser.add_argument("--result_path", required=True)
parser.add_argument("--model_level", required=True)
parser.add_argument("--ngf", type=int, default=32)
parser.add_argument("--n_residual_layers", type=int, default=3)
parser.add_argument("--ndf", type=int, default=32)
parser.add_argument("--num_D", type=int, default=3)
parser.add_argument("--n_layers_D", type=int, default=4)
parser.add_argument("--downsamp_factor", type=int, default=4)
parser.add_argument("--lambda_feat", type=float, default=10)
parser.add_argument("--cond_disc", action="store_true")
parser.add_argument("--data_path", default=None, type=Path)
parser.add_argument("--batch_size", type=int, default=16)
parser.add_argument("--epochs", type=int, default=3000)
parser.add_argument("--log_interval", type=int, default=100)
parser.add_argument("--save_interval", type=int, default=1000)
parser.add_argument("--n_test_samples", type=int, default=8)
args = parser.parse_args()
return args
# Generate and save samples, alignment, and webpage for visualization.
def generate(model, device, hps):
args = parse_args()
load_path = args.load_path
result_path = args.result_path
model_level = args.model_level
if model_level == "high":
code_level = 2
seq_len = 44032
level_s = 2
level_e = 3
if model_level == "low":
code_level = 1
seq_len = 44096
level_s = 1
level_e = 2
vqvae= make_vae_model(model, device, hps).cuda()
if model_level == "high":
encoder = vqEncoder_high().cuda()
vqvae.load_state_dict(t.load("./models/vqvae_high.pt"))
vqvae.eval()
encoder.eval()
if model_level == "low":
encoder = vqEncoder_low().cuda()
vqvae.load_state_dict(t.load("./models/vqvae_low.pt"))
vqvae.eval()
encoder.eval()
mencoder = motion_encoder().cuda()
mencoder.eval()
fft = Audio2Mel(n_mel_channels=128).cuda()
load_path_mencoder = os.path.join(load_path,'mencoder.pt')
load_path_encoder = os.path.join(load_path,'netG.pt')
mencoder.load_state_dict(t.load(load_path_mencoder))
encoder.load_state_dict(t.load(load_path_encoder))
# mencoder.load_state_dict(t.load("./logs/mencoder.pt"))
# encoder.load_state_dict(t.load("./logs/netG.pt"))
print("*******Finish model loading******")
#### creat data loader ####
# root = '/home/zhuye/musicgen'
va_set = TiktokDataset( audio_files = './dataset/tiktok_audio_test_segment.txt', video_files = './dataset/tiktok_video_test_segment.txt', motion_files = './dataset/tiktok_motion_test_segment.txt', augment=False)
va_loader = DataLoader(va_set, batch_size = 1)
print("*******Finish data loader*******")
#### generate samples ####
t.backends.cudnn.benchmark = True
for i, (a_t, v_t, m_t) in enumerate(va_loader):
a_t = a_t.float().cuda()
v_t = v_t.float().cuda()
m_t = m_t.float().cuda()
# gt samples and original samples
gt_xs, zs_code = vqvae._encode(a_t.transpose(1,2))
zs_middle = []
zs_middle.append(zs_code[code_level])
quantised_xs, out = vqvae._decode(zs_middle, start_level=level_s, end_level=level_e)
# original samples
audio = a_t.squeeze().cpu()
if not os.path.exists(result_path):
os.makedirs(result_path)
sample_original = 'original_' + str(i+1) + '.wav'
sample_vqvae = 'vqvae_'+ str(i+1) + '.wav'
sample_original = os.path.join(result_path,sample_original)
sample_vqvae = os.path.join(result_path,sample_vqvae)
sf.write(sample_original, audio.detach().cpu().numpy(), 22050)
sf.write(sample_vqvae, out.squeeze().detach().cpu().numpy(), 22050)
gt_code_error = F.l1_loss(gt_xs[code_level], quantised_xs[0])
gt_audio_error = F.l1_loss(audio[0:seq_len], out.squeeze().cpu())
## reconstructed samples
mx = mencoder(m_t)
fuse_x = t.cat((mx, v_t),2)
pred_xs = encoder(fuse_x)
xs_code = []
for w in range(3):
xs_code.append(pred_xs)
zs_pred = vqvae.bottleneck.encode(xs_code)
zs_pred_code = []
zs_pred_code.append(zs_pred[code_level])
xs_quantised_pred, pred_audio = vqvae._decode(zs_pred_code,start_level=level_s,end_level=level_e)
pred_code_error = F.l1_loss(pred_xs,quantised_xs[0] )
gen_audio_error = F.l1_loss(audio[0:seq_len], pred_audio.squeeze().cpu())
pred_audio = pred_audio.squeeze().detach().cpu().numpy()
pred_audio = nr.reduce_noise(y=pred_audio, sr=22050)
sample_generated = 'generated_'+ str(i+1) + '.wav'
sample_generated = os.path.join(result_path,sample_generated)
sf.write(sample_generated, pred_audio, 22050)
print("Generating testing sample:", i+1)
print("*******Finish generating samples*******")
def run(model, mode='ancestral', codes_file=None, audio_file=None, prompt_length_in_seconds=None, port=29500, **kwargs):
from jukebox.utils.dist_utils import setup_dist_from_mpi
rank, local_rank, device = setup_dist_from_mpi(port=port)
hps = Hyperparams(**kwargs)
# sample_hps = Hyperparams(dict(mode=mode, codes_file=codes_file, audio_file=audio_file, prompt_length_in_seconds=prompt_length_in_seconds))
with t.no_grad():
generate(model, device, hps)
if __name__ == '__main__':
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
fire.Fire(run)