init
Browse files- attach_speaker_embedding_s2s.py +7 -1
- main_s2s.sh +180 -127
attach_speaker_embedding_s2s.py
CHANGED
@@ -14,6 +14,8 @@ hf_dataset = f"seamless-align-{direction}"
|
|
14 |
dataset = load_dataset(f"{hf_org}/{hf_dataset}", f"subset_{dataset_id}", split="train")
|
15 |
audio_loader = Audio()
|
16 |
se_model = os.getenv("SE_MODEL", "metavoice")
|
|
|
|
|
17 |
if se_model == "metavoice":
|
18 |
from speaker_embedding_metavoice import MetaVoiceSE
|
19 |
speaker_embedder = MetaVoiceSE()
|
@@ -36,7 +38,11 @@ else:
|
|
36 |
def error_file(example):
|
37 |
for side in sides:
|
38 |
try:
|
39 |
-
audio_loader.decode_example(example[f"{side}.audio"])
|
|
|
|
|
|
|
|
|
40 |
except LibsndfileError:
|
41 |
return False
|
42 |
return True
|
|
|
14 |
dataset = load_dataset(f"{hf_org}/{hf_dataset}", f"subset_{dataset_id}", split="train")
|
15 |
audio_loader = Audio()
|
16 |
se_model = os.getenv("SE_MODEL", "metavoice")
|
17 |
+
max_seq_length = 10000000
|
18 |
+
|
19 |
if se_model == "metavoice":
|
20 |
from speaker_embedding_metavoice import MetaVoiceSE
|
21 |
speaker_embedder = MetaVoiceSE()
|
|
|
38 |
def error_file(example):
|
39 |
for side in sides:
|
40 |
try:
|
41 |
+
wav = audio_loader.decode_example(example[f"{side}.audio"])
|
42 |
+
if len(wav["array"]) == 0 or len(wav["array"]) > max_seq_length:
|
43 |
+
return False
|
44 |
+
except ValueError:
|
45 |
+
return False
|
46 |
except LibsndfileError:
|
47 |
return False
|
48 |
return True
|
main_s2s.sh
CHANGED
@@ -1,69 +1,44 @@
|
|
1 |
-
|
2 |
-
# enA-
|
3 |
-
|
4 |
# test
|
5 |
export DATASET_ID=test
|
6 |
-
export DIRECTION="enA-
|
7 |
export LINE_NO_START=0
|
8 |
export LINE_NO_END=10
|
9 |
python fetch_dataset_s2s.py
|
10 |
# main
|
11 |
-
for i in $(seq
|
12 |
do
|
13 |
export N_POOL=15
|
14 |
export DATASET_ID=${i}
|
15 |
-
export DIRECTION="enA-
|
16 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
17 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
18 |
echo ${LINE_NO_START}
|
19 |
python fetch_dataset_s2s.py
|
20 |
done
|
21 |
-
|
22 |
-
for i in $(seq 37 50);
|
23 |
do
|
24 |
export N_POOL=15
|
25 |
export DATASET_ID=${i}
|
26 |
-
export DIRECTION="enA-
|
27 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
28 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
29 |
echo ${LINE_NO_START}
|
30 |
-
python
|
31 |
done
|
32 |
-
for i in $(seq
|
33 |
do
|
34 |
export N_POOL=15
|
35 |
export DATASET_ID=${i}
|
36 |
-
export DIRECTION="enA-
|
37 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
38 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
39 |
echo ${LINE_NO_START}
|
40 |
-
python
|
41 |
-
done
|
42 |
-
# speaker embedding
|
43 |
-
for i in $(seq 1 20);
|
44 |
-
do
|
45 |
-
export DATASET_ID=${i}
|
46 |
-
export DIRECTION="enA-koA"
|
47 |
-
python attach_speaker_embedding_s2s.py
|
48 |
-
done
|
49 |
-
for i in $(seq 51 64);
|
50 |
-
do
|
51 |
-
export DATASET_ID=${i}
|
52 |
-
export DIRECTION="enA-koA"
|
53 |
-
python attach_speaker_embedding_s2s.py
|
54 |
done
|
55 |
-
|
56 |
-
######################
|
57 |
-
# enA-frA: 3_054_258 #
|
58 |
-
######################
|
59 |
-
# test
|
60 |
-
export DATASET_ID=test
|
61 |
-
export DIRECTION="enA-frA"
|
62 |
-
export LINE_NO_START=0
|
63 |
-
export LINE_NO_END=10
|
64 |
-
python fetch_dataset_s2s.py
|
65 |
-
# main
|
66 |
-
for i in $(seq 115 200);
|
67 |
do
|
68 |
export N_POOL=15
|
69 |
export DATASET_ID=${i}
|
@@ -73,7 +48,8 @@ do
|
|
73 |
echo ${LINE_NO_START}
|
74 |
python fetch_dataset_s2s.py
|
75 |
done
|
76 |
-
|
|
|
77 |
do
|
78 |
export N_POOL=15
|
79 |
export DATASET_ID=${i}
|
@@ -81,9 +57,9 @@ do
|
|
81 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
82 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
83 |
echo ${LINE_NO_START}
|
84 |
-
python
|
85 |
done
|
86 |
-
for i in $(seq
|
87 |
do
|
88 |
export N_POOL=15
|
89 |
export DATASET_ID=${i}
|
@@ -91,9 +67,9 @@ do
|
|
91 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
92 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
93 |
echo ${LINE_NO_START}
|
94 |
-
python
|
95 |
done
|
96 |
-
for i in $(seq
|
97 |
do
|
98 |
export N_POOL=15
|
99 |
export DATASET_ID=${i}
|
@@ -101,10 +77,9 @@ do
|
|
101 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
102 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
103 |
echo ${LINE_NO_START}
|
104 |
-
python
|
105 |
done
|
106 |
-
|
107 |
-
for i in $(seq 61 105);
|
108 |
do
|
109 |
export N_POOL=15
|
110 |
export DATASET_ID=${i}
|
@@ -115,14 +90,14 @@ do
|
|
115 |
python tokenize_dataset_s2s.py
|
116 |
done
|
117 |
# speaker embedding
|
118 |
-
for i in $(seq
|
119 |
do
|
120 |
export DATASET_ID=${i}
|
121 |
export DIRECTION="enA-frA"
|
122 |
python attach_speaker_embedding_s2s.py
|
123 |
done
|
124 |
# tokenize
|
125 |
-
for i in $(seq
|
126 |
do
|
127 |
export N_POOL=15
|
128 |
export DATASET_ID=${i}
|
@@ -161,7 +136,6 @@ do
|
|
161 |
echo ${LINE_NO_START}
|
162 |
python fetch_dataset_s2s.py
|
163 |
done
|
164 |
-
|
165 |
for i in $(seq 201 394);
|
166 |
do
|
167 |
export N_POOL=15
|
@@ -172,8 +146,7 @@ do
|
|
172 |
echo ${LINE_NO_START}
|
173 |
python fetch_dataset_s2s.py
|
174 |
done
|
175 |
-
|
176 |
-
for i in $(seq 51 104);
|
177 |
do
|
178 |
export N_POOL=15
|
179 |
export DATASET_ID=${i}
|
@@ -181,9 +154,20 @@ do
|
|
181 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
182 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
183 |
echo ${LINE_NO_START}
|
184 |
-
python
|
185 |
done
|
186 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
187 |
do
|
188 |
export N_POOL=15
|
189 |
export DATASET_ID=${i}
|
@@ -217,7 +201,7 @@ export LINE_NO_START=0
|
|
217 |
export LINE_NO_END=10
|
218 |
python fetch_dataset_s2s.py
|
219 |
# main
|
220 |
-
for i in $(seq
|
221 |
do
|
222 |
export N_POOL=15
|
223 |
export DATASET_ID=${i}
|
@@ -227,7 +211,7 @@ do
|
|
227 |
echo ${LINE_NO_START}
|
228 |
python fetch_dataset_s2s.py
|
229 |
done
|
230 |
-
for i in $(seq
|
231 |
do
|
232 |
export N_POOL=15
|
233 |
export DATASET_ID=${i}
|
@@ -269,25 +253,18 @@ do
|
|
269 |
python tokenize_dataset_s2s.py
|
270 |
done
|
271 |
# speaker embedding
|
272 |
-
for i in $(seq 1 50);
|
273 |
-
do
|
274 |
-
export DATASET_ID=${i}
|
275 |
-
export DIRECTION="enA-esA"
|
276 |
-
python attach_speaker_embedding_s2s.py
|
277 |
-
done
|
278 |
-
# tokenize
|
279 |
for i in $(seq 251 280);
|
280 |
do
|
281 |
-
export N_POOL=15
|
282 |
export DATASET_ID=${i}
|
283 |
export DIRECTION="enA-esA"
|
284 |
-
|
285 |
-
export LINE_NO_END=$((DATASET_ID * 2500))
|
286 |
-
echo ${LINE_NO_START}
|
287 |
-
python tokenize_dataset_s2s.py
|
288 |
done
|
289 |
-
|
290 |
-
|
|
|
|
|
|
|
|
|
291 |
do
|
292 |
export DATASET_ID=${i}
|
293 |
export DIRECTION="enA-esA"
|
@@ -296,100 +273,93 @@ done
|
|
296 |
|
297 |
|
298 |
|
299 |
-
|
300 |
####################
|
301 |
-
# enA-
|
302 |
####################
|
303 |
# test
|
304 |
export DATASET_ID=test
|
305 |
-
export DIRECTION="enA-
|
306 |
export LINE_NO_START=0
|
307 |
export LINE_NO_END=10
|
308 |
python fetch_dataset_s2s.py
|
309 |
# main
|
310 |
-
for i in $(seq 1
|
311 |
do
|
312 |
export N_POOL=15
|
313 |
export DATASET_ID=${i}
|
314 |
-
export DIRECTION="enA-
|
315 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
316 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
317 |
echo ${LINE_NO_START}
|
318 |
python fetch_dataset_s2s.py
|
319 |
done
|
320 |
# tokenize
|
321 |
-
for i in $(seq
|
322 |
do
|
|
|
323 |
export DATASET_ID=${i}
|
324 |
-
export DIRECTION="enA-
|
|
|
|
|
|
|
325 |
python tokenize_dataset_s2s.py
|
326 |
done
|
327 |
# speaker embedding
|
328 |
export SE_MODEL="metavoice"
|
329 |
-
export SE_MODEL="w2vbert-600m"
|
|
|
330 |
export SE_MODEL="hubert-xl"
|
331 |
-
|
332 |
-
for i in $(seq 17 91);
|
333 |
do
|
334 |
export DATASET_ID=${i}
|
335 |
-
export DIRECTION="enA-
|
336 |
python attach_speaker_embedding_s2s.py
|
337 |
done
|
338 |
|
339 |
-
export SE_MODEL="
|
340 |
-
|
341 |
-
export DIRECTION="enA-hiA"
|
342 |
-
python attach_speaker_embedding_s2s.py
|
343 |
-
####################
|
344 |
-
# enA-viA: 740_598 #
|
345 |
-
####################
|
346 |
-
# test
|
347 |
-
export DATASET_ID=test
|
348 |
-
export DIRECTION="enA-viA"
|
349 |
-
export LINE_NO_START=0
|
350 |
-
export LINE_NO_END=10
|
351 |
-
python fetch_dataset_s2s.py
|
352 |
-
# main
|
353 |
-
for i in $(seq 1 149);
|
354 |
do
|
355 |
-
export N_POOL=15
|
356 |
export DATASET_ID=${i}
|
357 |
-
export DIRECTION="enA-
|
358 |
-
|
359 |
-
export LINE_NO_END=$((DATASET_ID * 2500))
|
360 |
-
echo ${LINE_NO_START}
|
361 |
-
python fetch_dataset_s2s.py
|
362 |
done
|
363 |
-
|
364 |
-
for i in $(seq 1 149);
|
365 |
do
|
366 |
export DATASET_ID=${i}
|
367 |
-
export DIRECTION="enA-
|
368 |
-
python
|
369 |
done
|
370 |
-
|
371 |
-
|
372 |
-
export SE_MODEL="xlsr-2b"
|
373 |
-
export SE_MODEL="w2vbert-600m"
|
374 |
export SE_MODEL="hubert-xl"
|
375 |
-
for i in $(seq 1
|
376 |
do
|
377 |
export DATASET_ID=${i}
|
378 |
-
export DIRECTION="enA-
|
379 |
python attach_speaker_embedding_s2s.py
|
380 |
done
|
381 |
-
|
382 |
export SE_MODEL="w2vbert-600m"
|
383 |
-
|
384 |
-
python attach_speaker_embedding_s2s.py
|
385 |
-
|
386 |
-
export SE_MODEL="hubert-xl"
|
387 |
-
for i in 107 115 126;
|
388 |
do
|
389 |
export DATASET_ID=${i}
|
390 |
-
export DIRECTION="enA-
|
|
|
|
|
|
|
|
|
|
|
|
|
391 |
python attach_speaker_embedding_s2s.py
|
392 |
done
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
393 |
|
394 |
######################
|
395 |
# enA-zhA: 1_289_192 #
|
@@ -424,9 +394,9 @@ do
|
|
424 |
done
|
425 |
# speaker embedding
|
426 |
export SE_MODEL="metavoice"
|
427 |
-
#export SE_MODEL="xlsr-2b"
|
428 |
export SE_MODEL="w2vbert-600m"
|
429 |
-
export SE_MODEL="hubert-xl"
|
|
|
430 |
for i in $(seq 1 258);
|
431 |
do
|
432 |
export DATASET_ID=${i}
|
@@ -435,6 +405,28 @@ do
|
|
435 |
done
|
436 |
|
437 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
438 |
####################
|
439 |
# enA-jaA: 718_606 #
|
440 |
####################
|
@@ -466,7 +458,7 @@ done
|
|
466 |
export SE_MODEL="metavoice"
|
467 |
export SE_MODEL="w2vbert-600m"
|
468 |
export SE_MODEL="hubert-xl"
|
469 |
-
export SE_MODEL="xlsr-2b"
|
470 |
for i in $(seq 1 144);
|
471 |
do
|
472 |
export DATASET_ID=${i}
|
@@ -474,19 +466,80 @@ do
|
|
474 |
python attach_speaker_embedding_s2s.py
|
475 |
done
|
476 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
477 |
export SE_MODEL="w2vbert-600m"
|
478 |
-
|
|
|
|
|
479 |
do
|
480 |
export DATASET_ID=${i}
|
481 |
-
export DIRECTION="enA-
|
482 |
python attach_speaker_embedding_s2s.py
|
483 |
done
|
484 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
485 |
export SE_MODEL="hubert-xl"
|
486 |
-
|
|
|
487 |
do
|
488 |
export DATASET_ID=${i}
|
489 |
-
export DIRECTION="enA-
|
490 |
python attach_speaker_embedding_s2s.py
|
491 |
-
done
|
492 |
-
|
|
|
1 |
+
######################
|
2 |
+
# enA-frA: 3_054_258 #
|
3 |
+
######################
|
4 |
# test
|
5 |
export DATASET_ID=test
|
6 |
+
export DIRECTION="enA-frA"
|
7 |
export LINE_NO_START=0
|
8 |
export LINE_NO_END=10
|
9 |
python fetch_dataset_s2s.py
|
10 |
# main
|
11 |
+
for i in $(seq 115 200);
|
12 |
do
|
13 |
export N_POOL=15
|
14 |
export DATASET_ID=${i}
|
15 |
+
export DIRECTION="enA-frA"
|
16 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
17 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
18 |
echo ${LINE_NO_START}
|
19 |
python fetch_dataset_s2s.py
|
20 |
done
|
21 |
+
for i in $(seq 443 500);
|
|
|
22 |
do
|
23 |
export N_POOL=15
|
24 |
export DATASET_ID=${i}
|
25 |
+
export DIRECTION="enA-frA"
|
26 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
27 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
28 |
echo ${LINE_NO_START}
|
29 |
+
python fetch_dataset_s2s.py
|
30 |
done
|
31 |
+
for i in $(seq 519 550);
|
32 |
do
|
33 |
export N_POOL=15
|
34 |
export DATASET_ID=${i}
|
35 |
+
export DIRECTION="enA-frA"
|
36 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
37 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
38 |
echo ${LINE_NO_START}
|
39 |
+
python fetch_dataset_s2s.py
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
done
|
41 |
+
for i in $(seq 551 600);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
do
|
43 |
export N_POOL=15
|
44 |
export DATASET_ID=${i}
|
|
|
48 |
echo ${LINE_NO_START}
|
49 |
python fetch_dataset_s2s.py
|
50 |
done
|
51 |
+
# tokenize
|
52 |
+
for i in $(seq 61 145);
|
53 |
do
|
54 |
export N_POOL=15
|
55 |
export DATASET_ID=${i}
|
|
|
57 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
58 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
59 |
echo ${LINE_NO_START}
|
60 |
+
python tokenize_dataset_s2s.py
|
61 |
done
|
62 |
+
for i in $(seq 200 299);
|
63 |
do
|
64 |
export N_POOL=15
|
65 |
export DATASET_ID=${i}
|
|
|
67 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
68 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
69 |
echo ${LINE_NO_START}
|
70 |
+
python tokenize_dataset_s2s.py
|
71 |
done
|
72 |
+
for i in $(seq 426 450);
|
73 |
do
|
74 |
export N_POOL=15
|
75 |
export DATASET_ID=${i}
|
|
|
77 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
78 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
79 |
echo ${LINE_NO_START}
|
80 |
+
python tokenize_dataset_s2s.py
|
81 |
done
|
82 |
+
for i in $(seq 500 530);
|
|
|
83 |
do
|
84 |
export N_POOL=15
|
85 |
export DATASET_ID=${i}
|
|
|
90 |
python tokenize_dataset_s2s.py
|
91 |
done
|
92 |
# speaker embedding
|
93 |
+
for i in $(seq );
|
94 |
do
|
95 |
export DATASET_ID=${i}
|
96 |
export DIRECTION="enA-frA"
|
97 |
python attach_speaker_embedding_s2s.py
|
98 |
done
|
99 |
# tokenize
|
100 |
+
for i in $(seq 106 425);
|
101 |
do
|
102 |
export N_POOL=15
|
103 |
export DATASET_ID=${i}
|
|
|
136 |
echo ${LINE_NO_START}
|
137 |
python fetch_dataset_s2s.py
|
138 |
done
|
|
|
139 |
for i in $(seq 201 394);
|
140 |
do
|
141 |
export N_POOL=15
|
|
|
146 |
echo ${LINE_NO_START}
|
147 |
python fetch_dataset_s2s.py
|
148 |
done
|
149 |
+
for i in $(seq 73 91);
|
|
|
150 |
do
|
151 |
export N_POOL=15
|
152 |
export DATASET_ID=${i}
|
|
|
154 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
155 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
156 |
echo ${LINE_NO_START}
|
157 |
+
python fetch_dataset_s2s.py
|
158 |
done
|
159 |
+
# tokenize
|
160 |
+
#for i in $(seq 73 91);
|
161 |
+
#do
|
162 |
+
# export N_POOL=15
|
163 |
+
# export DATASET_ID=${i}
|
164 |
+
# export DIRECTION="deA-enA"
|
165 |
+
# export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
166 |
+
# export LINE_NO_END=$((DATASET_ID * 2500))
|
167 |
+
# echo ${LINE_NO_START}
|
168 |
+
# python tokenize_dataset_s2s.py
|
169 |
+
#done
|
170 |
+
for i in $(seq 105 180);
|
171 |
do
|
172 |
export N_POOL=15
|
173 |
export DATASET_ID=${i}
|
|
|
201 |
export LINE_NO_END=10
|
202 |
python fetch_dataset_s2s.py
|
203 |
# main
|
204 |
+
for i in $(seq 194 220);
|
205 |
do
|
206 |
export N_POOL=15
|
207 |
export DATASET_ID=${i}
|
|
|
211 |
echo ${LINE_NO_START}
|
212 |
python fetch_dataset_s2s.py
|
213 |
done
|
214 |
+
for i in $(seq 220 250);
|
215 |
do
|
216 |
export N_POOL=15
|
217 |
export DATASET_ID=${i}
|
|
|
253 |
python tokenize_dataset_s2s.py
|
254 |
done
|
255 |
# speaker embedding
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
256 |
for i in $(seq 251 280);
|
257 |
do
|
|
|
258 |
export DATASET_ID=${i}
|
259 |
export DIRECTION="enA-esA"
|
260 |
+
python attach_speaker_embedding_s2s.py
|
|
|
|
|
|
|
261 |
done
|
262 |
+
|
263 |
+
export SE_MODEL="w2vbert-600m" # running
|
264 |
+
export SE_MODEL="xlsr-2b" # running
|
265 |
+
export SE_MODEL="hubert-xl"
|
266 |
+
export SE_MODEL="metavoice"
|
267 |
+
for i in $(seq 26 125);
|
268 |
do
|
269 |
export DATASET_ID=${i}
|
270 |
export DIRECTION="enA-esA"
|
|
|
273 |
|
274 |
|
275 |
|
|
|
276 |
####################
|
277 |
+
# enA-koA: 511_358 #
|
278 |
####################
|
279 |
# test
|
280 |
export DATASET_ID=test
|
281 |
+
export DIRECTION="enA-koA"
|
282 |
export LINE_NO_START=0
|
283 |
export LINE_NO_END=10
|
284 |
python fetch_dataset_s2s.py
|
285 |
# main
|
286 |
+
for i in $(seq 1 100);
|
287 |
do
|
288 |
export N_POOL=15
|
289 |
export DATASET_ID=${i}
|
290 |
+
export DIRECTION="enA-koA"
|
291 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
292 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
293 |
echo ${LINE_NO_START}
|
294 |
python fetch_dataset_s2s.py
|
295 |
done
|
296 |
# tokenize
|
297 |
+
for i in $(seq 76 98);
|
298 |
do
|
299 |
+
export N_POOL=15
|
300 |
export DATASET_ID=${i}
|
301 |
+
export DIRECTION="enA-koA"
|
302 |
+
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
303 |
+
export LINE_NO_END=$((DATASET_ID * 2500))
|
304 |
+
echo ${LINE_NO_START}
|
305 |
python tokenize_dataset_s2s.py
|
306 |
done
|
307 |
# speaker embedding
|
308 |
export SE_MODEL="metavoice"
|
309 |
+
export SE_MODEL="w2vbert-600m" # running
|
310 |
+
export SE_MODEL="xlsr-2b" # running
|
311 |
export SE_MODEL="hubert-xl"
|
312 |
+
for i in $(seq 1 90);
|
|
|
313 |
do
|
314 |
export DATASET_ID=${i}
|
315 |
+
export DIRECTION="enA-koA"
|
316 |
python attach_speaker_embedding_s2s.py
|
317 |
done
|
318 |
|
319 |
+
export SE_MODEL="metavoice"
|
320 |
+
for i in 65 50;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
321 |
do
|
|
|
322 |
export DATASET_ID=${i}
|
323 |
+
export DIRECTION="enA-koA"
|
324 |
+
python attach_speaker_embedding_s2s.py
|
|
|
|
|
|
|
325 |
done
|
326 |
+
for i in $(seq 90 98);
|
|
|
327 |
do
|
328 |
export DATASET_ID=${i}
|
329 |
+
export DIRECTION="enA-koA"
|
330 |
+
python attach_speaker_embedding_s2s.py
|
331 |
done
|
332 |
+
|
333 |
+
# todo
|
|
|
|
|
334 |
export SE_MODEL="hubert-xl"
|
335 |
+
for i in $(seq 1 98);
|
336 |
do
|
337 |
export DATASET_ID=${i}
|
338 |
+
export DIRECTION="enA-koA"
|
339 |
python attach_speaker_embedding_s2s.py
|
340 |
done
|
|
|
341 |
export SE_MODEL="w2vbert-600m"
|
342 |
+
for i in $(seq 90 98);
|
|
|
|
|
|
|
|
|
343 |
do
|
344 |
export DATASET_ID=${i}
|
345 |
+
export DIRECTION="enA-koA"
|
346 |
+
python attach_speaker_embedding_s2s.py
|
347 |
+
done
|
348 |
+
for i in 34 49 50 87 88;
|
349 |
+
do
|
350 |
+
export DATASET_ID=${i}
|
351 |
+
export DIRECTION="enA-koA"
|
352 |
python attach_speaker_embedding_s2s.py
|
353 |
done
|
354 |
+
export SE_MODEL="xlsr-2b"
|
355 |
+
for i in $(seq 90 98);
|
356 |
+
do
|
357 |
+
export DATASET_ID=${i}
|
358 |
+
export DIRECTION="enA-koA"
|
359 |
+
python attach_speaker_embedding_s2s.py
|
360 |
+
done
|
361 |
+
|
362 |
+
|
363 |
|
364 |
######################
|
365 |
# enA-zhA: 1_289_192 #
|
|
|
394 |
done
|
395 |
# speaker embedding
|
396 |
export SE_MODEL="metavoice"
|
|
|
397 |
export SE_MODEL="w2vbert-600m"
|
398 |
+
export SE_MODEL="hubert-xl" # running
|
399 |
+
export SE_MODEL="xlsr-2b" # running
|
400 |
for i in $(seq 1 258);
|
401 |
do
|
402 |
export DATASET_ID=${i}
|
|
|
405 |
done
|
406 |
|
407 |
|
408 |
+
export SE_MODEL="hubert-xl" # running
|
409 |
+
for i in 165 167;
|
410 |
+
do
|
411 |
+
export DATASET_ID=${i}
|
412 |
+
export DIRECTION="enA-zhA"
|
413 |
+
python attach_speaker_embedding_s2s.py
|
414 |
+
done
|
415 |
+
|
416 |
+
|
417 |
+
export SE_MODEL="w2vbert-600m" # running
|
418 |
+
for i in 142 165 167 207 253;
|
419 |
+
do
|
420 |
+
export DATASET_ID=${i}
|
421 |
+
export DIRECTION="enA-zhA"
|
422 |
+
python attach_speaker_embedding_s2s.py
|
423 |
+
done
|
424 |
+
|
425 |
+
|
426 |
+
|
427 |
+
|
428 |
+
|
429 |
+
|
430 |
####################
|
431 |
# enA-jaA: 718_606 #
|
432 |
####################
|
|
|
458 |
export SE_MODEL="metavoice"
|
459 |
export SE_MODEL="w2vbert-600m"
|
460 |
export SE_MODEL="hubert-xl"
|
461 |
+
export SE_MODEL="xlsr-2b" # running
|
462 |
for i in $(seq 1 144);
|
463 |
do
|
464 |
export DATASET_ID=${i}
|
|
|
466 |
python attach_speaker_embedding_s2s.py
|
467 |
done
|
468 |
|
469 |
+
####################
|
470 |
+
# enA-viA: 740_598 #
|
471 |
+
####################
|
472 |
+
# test
|
473 |
+
export DATASET_ID=test
|
474 |
+
export DIRECTION="enA-viA"
|
475 |
+
export LINE_NO_START=0
|
476 |
+
export LINE_NO_END=10
|
477 |
+
python fetch_dataset_s2s.py
|
478 |
+
# main
|
479 |
+
for i in $(seq 1 149);
|
480 |
+
do
|
481 |
+
export N_POOL=15
|
482 |
+
export DATASET_ID=${i}
|
483 |
+
export DIRECTION="enA-viA"
|
484 |
+
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
485 |
+
export LINE_NO_END=$((DATASET_ID * 2500))
|
486 |
+
echo ${LINE_NO_START}
|
487 |
+
python fetch_dataset_s2s.py
|
488 |
+
done
|
489 |
+
# tokenize
|
490 |
+
for i in $(seq 1 149);
|
491 |
+
do
|
492 |
+
export DATASET_ID=${i}
|
493 |
+
export DIRECTION="enA-viA"
|
494 |
+
python tokenize_dataset_s2s.py
|
495 |
+
done
|
496 |
+
# speaker embedding
|
497 |
+
export SE_MODEL="metavoice"
|
498 |
export SE_MODEL="w2vbert-600m"
|
499 |
+
export SE_MODEL="hubert-xl"
|
500 |
+
export SE_MODEL="xlsr-2b"
|
501 |
+
for i in $(seq 1 149);
|
502 |
do
|
503 |
export DATASET_ID=${i}
|
504 |
+
export DIRECTION="enA-viA"
|
505 |
python attach_speaker_embedding_s2s.py
|
506 |
done
|
507 |
|
508 |
+
####################
|
509 |
+
# enA-hiA: 454_942 #
|
510 |
+
####################
|
511 |
+
# test
|
512 |
+
export DATASET_ID=test
|
513 |
+
export DIRECTION="enA-hiA"
|
514 |
+
export LINE_NO_START=0
|
515 |
+
export LINE_NO_END=10
|
516 |
+
python fetch_dataset_s2s.py
|
517 |
+
# main
|
518 |
+
for i in $(seq 1 91);
|
519 |
+
do
|
520 |
+
export N_POOL=15
|
521 |
+
export DATASET_ID=${i}
|
522 |
+
export DIRECTION="enA-hiA"
|
523 |
+
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
524 |
+
export LINE_NO_END=$((DATASET_ID * 2500))
|
525 |
+
echo ${LINE_NO_START}
|
526 |
+
python fetch_dataset_s2s.py
|
527 |
+
done
|
528 |
+
# tokenize
|
529 |
+
for i in $(seq 1 91);
|
530 |
+
do
|
531 |
+
export DATASET_ID=${i}
|
532 |
+
export DIRECTION="enA-hiA"
|
533 |
+
python tokenize_dataset_s2s.py
|
534 |
+
done
|
535 |
+
# speaker embedding
|
536 |
+
export SE_MODEL="metavoice"
|
537 |
+
export SE_MODEL="w2vbert-600m"
|
538 |
export SE_MODEL="hubert-xl"
|
539 |
+
export SE_MODEL="xlsr-2b" # running
|
540 |
+
for i in $(seq 1 91);
|
541 |
do
|
542 |
export DATASET_ID=${i}
|
543 |
+
export DIRECTION="enA-hiA"
|
544 |
python attach_speaker_embedding_s2s.py
|
545 |
+
done
|
|