-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathindex.html
360 lines (353 loc) · 22.7 KB
/
index.html
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
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>Audio samples from "HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis"</title>
<meta charset="UTF-8">
<title>HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis</title>
<style>
red {color: red}
audio {width: 250px}
</style>
</head>
<body>
<article>
<header>
<h1>Audio samples from "HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis"</h1>
</header>
</article>
<!-- <p><b>Paper:</b> <a href="">arXiv</a></p>-->
<!-- <p><b>Authors</b>: </p>-->
<div>
<b>Abstract:</b>
Several recent work on speech synthesis have employed generative adversarial networks (GANs) to produce raw waveforms.
Although such methods improve the sampling efficiency and memory usage,
their sample quality has not yet reached that of autoregressive and flow-based generative models.
In this work, we propose HiFi-GAN, which achieves both efficient and high-fidelity speech synthesis.
As speech audio consists of sinusoidal signals with various periods, we demonstrate that modeling periodic patterns of an audio is crucial for enhancing sample quality.
A subjective human evaluation (mean opinion score, MOS) of a single speaker dataset indicates
that our proposed method demonstrates similarity to human quality while generating 22.05 kHz high-fidelity audio 167.9 times faster
than real-time on a single V100 GPU. We further show the generality of HiFi-GAN to the mel-spectrogram inversion of unseen speakers and end-to-end speech synthesis.
Finally, a small footprint version of HiFi-GAN generates samples 13.4 times faster than real-time on CPU with comparable quality to an autoregressive counterpart.
<p></p>
</div>
<p>For more details of our work, please refer to the <a href="https://arxiv.org/abs/2010.05646">paper</a>.<br/>
Our implementation is available in the <a href="https://github.com/jik876/hifi-gan">github repository</a>.<br/><br/>
</p>
<p class="toc_title">Contents</p>
<div id="toc_container">
<ul>
<li><a href="#ss">Single Speaker (LJ Speech Dataset)</a></li>
<li><a href="#ms">Unseen Speakers (VCTK Dataset)</a></li>
<li><a href="#e2e">End-to-end Speech Synthesis (LJ Speech Dataset)</a></li>
<li><a href="#as">Ablation Studies (LJ Speech Dataset)</a></li>
</ul>
</div>
<p> </p>
<div>
<a name="ss"><h2>Single Speaker (LJ Speech Dataset)</h2></a>
<hr>
<table>
<tbody>
<tr>
<td nowrap width="160">Ground Truth</td>
<td><audio controls="" preload="none"><source src="samples/ss/GT/ss_gt_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/GT/ss_gt_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/GT/ss_gt_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/GT/ss_gt_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/GT/ss_gt_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>WaveNet (MoL)
<td><audio controls="" preload="none"><source src="samples/ss/WN/ss_wn_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WN/ss_wn_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WN/ss_wn_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WN/ss_wn_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WN/ss_wn_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>WaveGlow
<td><audio controls="" preload="none"><source src="samples/ss/WG/ss_wg_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WG/ss_wg_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WG/ss_wg_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WG/ss_wg_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/WG/ss_wg_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>MelGAN
<td><audio controls="" preload="none"><source src="samples/ss/MG/ss_mg_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/MG/ss_mg_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/MG/ss_mg_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/MG/ss_mg_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/MG/ss_mg_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V1 (ours)
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V1/ss_hgan_v1_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V1/ss_hgan_v1_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V1/ss_hgan_v1_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V1/ss_hgan_v1_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V1/ss_hgan_v1_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V2 (ours)
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V2/ss_hgan_v2_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V2/ss_hgan_v2_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V2/ss_hgan_v2_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V2/ss_hgan_v2_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V2/ss_hgan_v2_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V3 (ours)
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V3/ss_hgan_v3_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V3/ss_hgan_v3_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V3/ss_hgan_v3_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V3/ss_hgan_v3_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ss/HGAN_V3/ss_hgan_v3_5.wav"></audio></td>
</tr>
</tbody>
</table>
</div>
<br><br>
<div>
<a name="ms"><h2>Unseen Speakers (VCTK Dataset)</h2></a>
<hr>
<table>
<tbody>
<tr>
<td nowrap width="160">Ground Truth</td>
<td><audio controls="" preload="none"><source src="samples/ms/GT/ms_gt_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/GT/ms_gt_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/GT/ms_gt_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/GT/ms_gt_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/GT/ms_gt_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>WaveNet (MoL)
<td><audio controls="" preload="none"><source src="samples/ms/WN/ms_wn_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WN/ms_wn_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WN/ms_wn_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WN/ms_wn_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WN/ms_wn_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>WaveGlow
<td><audio controls="" preload="none"><source src="samples/ms/WG/ms_wg_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WG/ms_wg_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WG/ms_wg_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WG/ms_wg_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/WG/ms_wg_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>MelGAN
<td><audio controls="" preload="none"><source src="samples/ms/MG/ms_mg_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/MG/ms_mg_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/MG/ms_mg_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/MG/ms_mg_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/MG/ms_mg_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V1 (ours)
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V1/ms_hgan_v1_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V1/ms_hgan_v1_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V1/ms_hgan_v1_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V1/ms_hgan_v1_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V1/ms_hgan_v1_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V2 (ours)
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V3 (ours)
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/ms/HGAN_V2/ms_hgan_v2_5.wav"></audio></td>
</tr>
</tbody>
</table>
</div>
<br><br>
<div>
<a name="e2e"><h2>End-to-end Speech Synthesis (LJ Speech Dataset)</h2></a>
<hr>
<table>
<tbody>
<tr>
<td nowrap width="160">Ground Truth</td>
<td><audio controls="" preload="none"><source src="samples/e2e/GT/e2e_gt_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/GT/e2e_gt_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/GT/e2e_gt_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/GT/e2e_gt_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/GT/e2e_gt_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>WaveGlow<br>(fine-tuned)
<td><audio controls="" preload="none"><source src="samples/e2e/WG_FT/e2e_wg_ft_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG_FT/e2e_wg_ft_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG_FT/e2e_wg_ft_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG_FT/e2e_wg_ft_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG_FT/e2e_wg_ft_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V1 (ours)<br>(fine-tuned)
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1_FT/e2e_hgan_v1_ft_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1_FT/e2e_hgan_v1_ft_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1_FT/e2e_hgan_v1_ft_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1_FT/e2e_hgan_v1_ft_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1_FT/e2e_hgan_v1_ft_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V2 (ours)<br>(fine-tuned)
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2_FT/e2e_hgan_v2_ft_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2_FT/e2e_hgan_v2_ft_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2_FT/e2e_hgan_v2_ft_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2_FT/e2e_hgan_v2_ft_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2_FT/e2e_hgan_v2_ft_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V3 (ours)<br>(fine-tuned)
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3_FT/e2e_hgan_v3_ft_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3_FT/e2e_hgan_v3_ft_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3_FT/e2e_hgan_v3_ft_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3_FT/e2e_hgan_v3_ft_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3_FT/e2e_hgan_v3_ft_5.wav"></audio></td>
</tr>
<tr>
<td colspan="6"><hr></td>
</tr>
<tr>
<td nowrap>WaveGlow<br>(w/o fine-tuning)
<td><audio controls="" preload="none"><source src="samples/e2e/WG/e2e_wg_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG/e2e_wg_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG/e2e_wg_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG/e2e_wg_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/WG/e2e_wg_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V1 (ours)<br>(w/o fine-tuning)
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1/e2e_hgan_v1_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1/e2e_hgan_v1_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1/e2e_hgan_v1_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1/e2e_hgan_v1_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V1/e2e_hgan_v1_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V2 (ours)<br>(w/o fine-tuning)
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2/e2e_hgan_v2_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2/e2e_hgan_v2_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2/e2e_hgan_v2_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2/e2e_hgan_v2_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V2/e2e_hgan_v2_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>HiFi-GAN V3 (ours)<br>(w/o fine-tuning)
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3/e2e_hgan_v3_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3/e2e_hgan_v3_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3/e2e_hgan_v3_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3/e2e_hgan_v3_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/e2e/HGAN_V3/e2e_hgan_v3_5.wav"></audio></td>
</tr>
</tbody>
</table>
</div>
<br><br>
<div>
<a name="as"><h2>Ablation Studies (LJ Speech Dataset)</h2></a>
<hr>
<table>
<tbody>
<tr>
<td nowrap>baseline
<td><audio controls="" preload="none"><source src="samples/as/baseline/as_baseline_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/baseline/as_baseline_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/baseline/as_baseline_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/baseline/as_baseline_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/baseline/as_baseline_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>w/o MPD
<td><audio controls="" preload="none"><source src="samples/as/wo_MPD/as_wo_mpd_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MPD/as_wo_mpd_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MPD/as_wo_mpd_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MPD/as_wo_mpd_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MPD/as_wo_mpd_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>w/o MSD
<td><audio controls="" preload="none"><source src="samples/as/wo_MSD/as_wo_msd_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MSD/as_wo_msd_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MSD/as_wo_msd_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MSD/as_wo_msd_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MSD/as_wo_msd_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>w/o MRF
<td><audio controls="" preload="none"><source src="samples/as/wo_MRF/as_wo_mrf_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MRF/as_wo_mrf_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MRF/as_wo_mrf_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MRF/as_wo_mrf_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_MRF/as_wo_mrf_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>w/o Mel-Spectrogram Loss
<td><audio controls="" preload="none"><source src="samples/as/wo_L1/as_wo_l1_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_L1/as_wo_l1_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_L1/as_wo_l1_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_L1/as_wo_l1_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as/wo_L1/as_wo_l1_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>MPD <i>p</i>=[2,4,8,16,32]
<td><audio controls="" preload="none"><source src="samples/as3/pr/as_pr_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as3/pr/as_pr_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as3/pr/as_pr_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as3/pr/as_pr_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as3/pr/as_pr_5.wav"></audio></td>
</tr>
</tbody>
</table>
<hr>
<table>
<tbody>
<tr>
<td nowrap width="198">MelGAN with MPD</td>
<td><audio controls="" preload="none"><source src="samples/as2/MG_MPD/as_mg_mpd_500_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG_MPD/as_mg_mpd_500_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG_MPD/as_mg_mpd_500_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG_MPD/as_mg_mpd_500_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG_MPD/as_mg_mpd_500_5.wav"></audio></td>
</tr>
<tr>
<td nowrap>MelGAN</td>
<td><audio controls="" preload="none"><source src="samples/as2/MG/as_mg_500_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG/as_mg_500_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG/as_mg_500_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG/as_mg_500_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/MG/as_mg_500_5.wav"></audio></td>
</tr>
</tbody>
</table>
<hr>
<table>
<tbody>
<tr>
<td nowrap width="198">HiFi-GAN V1 (500k step)</td>
<td><audio controls="" preload="none"><source src="samples/as2/V1_500/as_v1_500_1.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/V1_500/as_v1_500_2.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/V1_500/as_v1_500_3.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/V1_500/as_v1_500_4.wav"></audio></td>
<td><audio controls="" preload="none"><source src="samples/as2/V1_500/as_v1_500_5.wav"></audio></td>
</tr>
</tbody>
</table>
</div>
<br><br>
</body>
</html>