add streamable seamlessm4t
Browse files- fetch_dataset_s2s.py +5 -1
- main_s2s.sh +128 -87
fetch_dataset_s2s.py
CHANGED
@@ -49,7 +49,11 @@ min_seq_length = 50000
|
|
49 |
|
50 |
def wget(url: str, output_file: Optional[str] = None):
|
51 |
os.makedirs(os.path.dirname(output_file), exist_ok=True)
|
52 |
-
subprocess.
|
|
|
|
|
|
|
|
|
53 |
if not os.path.exists(output_file):
|
54 |
return False
|
55 |
if output_file.endswith('.tar.gz') or output_file.endswith('.tgz') or output_file.endswith('.tar'):
|
|
|
49 |
|
50 |
def wget(url: str, output_file: Optional[str] = None):
|
51 |
os.makedirs(os.path.dirname(output_file), exist_ok=True)
|
52 |
+
subprocess.check_output(
|
53 |
+
["wget", url, "-O", output_file, "--tries", wget_max_retry, "--timeout", wget_timeout],
|
54 |
+
stderr=subprocess.STDOUT,
|
55 |
+
timeout=int(60 * 5) # seconds (if it takes more than 5 min, skip it)
|
56 |
+
)
|
57 |
if not os.path.exists(output_file):
|
58 |
return False
|
59 |
if output_file.endswith('.tar.gz') or output_file.endswith('.tgz') or output_file.endswith('.tar'):
|
main_s2s.sh
CHANGED
@@ -8,7 +8,7 @@ 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}
|
@@ -18,7 +18,7 @@ do
|
|
18 |
echo ${LINE_NO_START}
|
19 |
python fetch_dataset_s2s.py
|
20 |
done
|
21 |
-
for i in $(seq
|
22 |
do
|
23 |
export N_POOL=15
|
24 |
export DATASET_ID=${i}
|
@@ -28,7 +28,7 @@ do
|
|
28 |
echo ${LINE_NO_START}
|
29 |
python fetch_dataset_s2s.py
|
30 |
done
|
31 |
-
for i in $(seq
|
32 |
do
|
33 |
export N_POOL=15
|
34 |
export DATASET_ID=${i}
|
@@ -38,18 +38,7 @@ do
|
|
38 |
echo ${LINE_NO_START}
|
39 |
python fetch_dataset_s2s.py
|
40 |
done
|
41 |
-
|
42 |
-
for i in $(seq 61 145);
|
43 |
-
do
|
44 |
-
export N_POOL=15
|
45 |
-
export DATASET_ID=${i}
|
46 |
-
export DIRECTION="enA-frA"
|
47 |
-
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
48 |
-
export LINE_NO_END=$((DATASET_ID * 2500))
|
49 |
-
echo ${LINE_NO_START}
|
50 |
-
python tokenize_dataset_s2s.py
|
51 |
-
done
|
52 |
-
for i in $(seq 200 299);
|
53 |
do
|
54 |
export N_POOL=15
|
55 |
export DATASET_ID=${i}
|
@@ -57,19 +46,10 @@ do
|
|
57 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
58 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
59 |
echo ${LINE_NO_START}
|
60 |
-
python
|
61 |
-
done
|
62 |
-
for i in $(seq 426 450);
|
63 |
-
do
|
64 |
-
export N_POOL=15
|
65 |
-
export DATASET_ID=${i}
|
66 |
-
export DIRECTION="enA-frA"
|
67 |
-
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
68 |
-
export LINE_NO_END=$((DATASET_ID * 2500))
|
69 |
-
echo ${LINE_NO_START}
|
70 |
-
python tokenize_dataset_s2s.py
|
71 |
done
|
72 |
-
|
|
|
73 |
do
|
74 |
export N_POOL=15
|
75 |
export DATASET_ID=${i}
|
@@ -81,34 +61,10 @@ do
|
|
81 |
done
|
82 |
# speaker embedding
|
83 |
export SE_MODEL="metavoice"
|
84 |
-
export SE_MODEL="hubert-xl"
|
85 |
-
export SE_MODEL="xlsr-2b"
|
86 |
-
export SE_MODEL="w2vbert-600m"
|
87 |
-
for i in $(seq 1
|
88 |
-
do
|
89 |
-
export DATASET_ID=${i}
|
90 |
-
export DIRECTION="enA-frA"
|
91 |
-
python attach_speaker_embedding_s2s.py
|
92 |
-
done
|
93 |
-
for i in $(seq 200 240);
|
94 |
-
do
|
95 |
-
export DATASET_ID=${i}
|
96 |
-
export DIRECTION="enA-frA"
|
97 |
-
python attach_speaker_embedding_s2s.py
|
98 |
-
done
|
99 |
-
for i in $(seq 300 380);
|
100 |
-
do
|
101 |
-
export DATASET_ID=${i}
|
102 |
-
export DIRECTION="enA-frA"
|
103 |
-
python attach_speaker_embedding_s2s.py
|
104 |
-
done
|
105 |
-
for i in $(seq 400 460);
|
106 |
-
do
|
107 |
-
export DATASET_ID=${i}
|
108 |
-
export DIRECTION="enA-frA"
|
109 |
-
python attach_speaker_embedding_s2s.py
|
110 |
-
done
|
111 |
-
for i in $(seq 500 570);
|
112 |
do
|
113 |
export DATASET_ID=${i}
|
114 |
export DIRECTION="enA-frA"
|
@@ -126,14 +82,7 @@ export LINE_NO_START=0
|
|
126 |
export LINE_NO_END=10
|
127 |
python fetch_dataset_s2s.py
|
128 |
# main
|
129 |
-
|
130 |
-
export DATASET_ID=54
|
131 |
-
export DIRECTION="enA-esA"
|
132 |
-
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
133 |
-
export LINE_NO_END=$((DATASET_ID * 2500))
|
134 |
-
echo ${LINE_NO_START}
|
135 |
-
python fetch_dataset_s2s.py
|
136 |
-
for i in $(seq 210 219);
|
137 |
do
|
138 |
export N_POOL=15
|
139 |
export DATASET_ID=${i}
|
@@ -143,7 +92,7 @@ do
|
|
143 |
echo ${LINE_NO_START}
|
144 |
python fetch_dataset_s2s.py
|
145 |
done
|
146 |
-
for i in $(seq
|
147 |
do
|
148 |
export N_POOL=15
|
149 |
export DATASET_ID=${i}
|
@@ -153,7 +102,7 @@ do
|
|
153 |
echo ${LINE_NO_START}
|
154 |
python fetch_dataset_s2s.py
|
155 |
done
|
156 |
-
for i in $(seq
|
157 |
do
|
158 |
export N_POOL=15
|
159 |
export DATASET_ID=${i}
|
@@ -163,7 +112,7 @@ do
|
|
163 |
echo ${LINE_NO_START}
|
164 |
python fetch_dataset_s2s.py
|
165 |
done
|
166 |
-
for i in $(seq
|
167 |
do
|
168 |
export N_POOL=15
|
169 |
export DATASET_ID=${i}
|
@@ -216,7 +165,7 @@ export LINE_NO_START=0
|
|
216 |
export LINE_NO_END=10
|
217 |
python fetch_dataset_s2s.py
|
218 |
# main
|
219 |
-
for i in $(seq
|
220 |
do
|
221 |
export N_POOL=15
|
222 |
export DATASET_ID=${i}
|
@@ -226,6 +175,7 @@ do
|
|
226 |
echo ${LINE_NO_START}
|
227 |
python fetch_dataset_s2s.py
|
228 |
done
|
|
|
229 |
for i in $(seq 321 353);
|
230 |
do
|
231 |
export N_POOL=15
|
@@ -234,39 +184,124 @@ do
|
|
234 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
235 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
236 |
echo ${LINE_NO_START}
|
237 |
-
python
|
238 |
done
|
239 |
-
|
240 |
-
|
|
|
241 |
do
|
242 |
-
export N_POOL=15
|
243 |
export DATASET_ID=${i}
|
244 |
export DIRECTION="deA-enA"
|
245 |
-
|
246 |
-
export LINE_NO_END=$((DATASET_ID * 2500))
|
247 |
-
echo ${LINE_NO_START}
|
248 |
-
python tokenize_dataset_s2s.py
|
249 |
done
|
250 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
251 |
do
|
252 |
-
export N_POOL=15
|
253 |
export DATASET_ID=${i}
|
254 |
-
export DIRECTION="
|
255 |
-
|
256 |
-
export LINE_NO_END=$((DATASET_ID * 2500))
|
257 |
-
echo ${LINE_NO_START}
|
258 |
-
python tokenize_dataset_s2s.py
|
259 |
done
|
260 |
-
|
261 |
-
for i in $(seq 28 104);
|
262 |
do
|
263 |
export DATASET_ID=${i}
|
264 |
-
export DIRECTION="
|
265 |
python attach_speaker_embedding_s2s.py
|
266 |
done
|
267 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
268 |
export SE_MODEL="w2vbert-600m"
|
269 |
-
for i in $(seq
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
270 |
do
|
271 |
export DATASET_ID=${i}
|
272 |
export DIRECTION="deA-enA"
|
@@ -278,6 +313,12 @@ done
|
|
278 |
|
279 |
|
280 |
|
|
|
|
|
|
|
|
|
|
|
|
|
281 |
####################
|
282 |
# enA-koA: 511_358 #
|
283 |
####################
|
|
|
8 |
export LINE_NO_END=10
|
9 |
python fetch_dataset_s2s.py
|
10 |
# main
|
11 |
+
for i in $(seq 181 200);
|
12 |
do
|
13 |
export N_POOL=15
|
14 |
export DATASET_ID=${i}
|
|
|
18 |
echo ${LINE_NO_START}
|
19 |
python fetch_dataset_s2s.py
|
20 |
done
|
21 |
+
for i in $(seq 256 280);
|
22 |
do
|
23 |
export N_POOL=15
|
24 |
export DATASET_ID=${i}
|
|
|
28 |
echo ${LINE_NO_START}
|
29 |
python fetch_dataset_s2s.py
|
30 |
done
|
31 |
+
for i in $(seq 281 299);
|
32 |
do
|
33 |
export N_POOL=15
|
34 |
export DATASET_ID=${i}
|
|
|
38 |
echo ${LINE_NO_START}
|
39 |
python fetch_dataset_s2s.py
|
40 |
done
|
41 |
+
for i in $(seq 1 600);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
do
|
43 |
export N_POOL=15
|
44 |
export DATASET_ID=${i}
|
|
|
46 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
47 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
48 |
echo ${LINE_NO_START}
|
49 |
+
python fetch_dataset_s2s.py
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
done
|
51 |
+
# tokenize
|
52 |
+
for i in $(seq 1 600);
|
53 |
do
|
54 |
export N_POOL=15
|
55 |
export DATASET_ID=${i}
|
|
|
61 |
done
|
62 |
# speaker embedding
|
63 |
export SE_MODEL="metavoice"
|
64 |
+
export SE_MODEL="hubert-xl"
|
65 |
+
export SE_MODEL="xlsr-2b"
|
66 |
+
export SE_MODEL="w2vbert-600m"
|
67 |
+
for i in $(seq 1 600);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
68 |
do
|
69 |
export DATASET_ID=${i}
|
70 |
export DIRECTION="enA-frA"
|
|
|
82 |
export LINE_NO_END=10
|
83 |
python fetch_dataset_s2s.py
|
84 |
# main
|
85 |
+
for i in $(seq 345 420);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
86 |
do
|
87 |
export N_POOL=15
|
88 |
export DATASET_ID=${i}
|
|
|
92 |
echo ${LINE_NO_START}
|
93 |
python fetch_dataset_s2s.py
|
94 |
done
|
95 |
+
for i in $(seq 390 420);
|
96 |
do
|
97 |
export N_POOL=15
|
98 |
export DATASET_ID=${i}
|
|
|
102 |
echo ${LINE_NO_START}
|
103 |
python fetch_dataset_s2s.py
|
104 |
done
|
105 |
+
for i in $(seq 451 500);
|
106 |
do
|
107 |
export N_POOL=15
|
108 |
export DATASET_ID=${i}
|
|
|
112 |
echo ${LINE_NO_START}
|
113 |
python fetch_dataset_s2s.py
|
114 |
done
|
115 |
+
for i in $(seq 500 532);
|
116 |
do
|
117 |
export N_POOL=15
|
118 |
export DATASET_ID=${i}
|
|
|
165 |
export LINE_NO_END=10
|
166 |
python fetch_dataset_s2s.py
|
167 |
# main
|
168 |
+
for i in $(seq 1 394);
|
169 |
do
|
170 |
export N_POOL=15
|
171 |
export DATASET_ID=${i}
|
|
|
175 |
echo ${LINE_NO_START}
|
176 |
python fetch_dataset_s2s.py
|
177 |
done
|
178 |
+
# tokenize
|
179 |
for i in $(seq 321 353);
|
180 |
do
|
181 |
export N_POOL=15
|
|
|
184 |
export LINE_NO_START=$(((DATASET_ID-1) * 2500))
|
185 |
export LINE_NO_END=$((DATASET_ID * 2500))
|
186 |
echo ${LINE_NO_START}
|
187 |
+
python tokenize_dataset_s2s.py
|
188 |
done
|
189 |
+
|
190 |
+
# speaker embedding
|
191 |
+
for i in $(seq 28 104);
|
192 |
do
|
|
|
193 |
export DATASET_ID=${i}
|
194 |
export DIRECTION="deA-enA"
|
195 |
+
python attach_speaker_embedding_s2s.py
|
|
|
|
|
|
|
196 |
done
|
197 |
+
# speaker embedding
|
198 |
+
|
199 |
+
|
200 |
+
# running on kotoba_2
|
201 |
+
#TODO1
|
202 |
+
export SE_MODEL="xlsr-2b"
|
203 |
+
for i in 1 2 4 5 6 10 41 50 55 340 377 379;
|
204 |
do
|
|
|
205 |
export DATASET_ID=${i}
|
206 |
+
export DIRECTION="enA-frA"
|
207 |
+
python attach_speaker_embedding_s2s.py
|
|
|
|
|
|
|
208 |
done
|
209 |
+
for i in $(seq 73 299);
|
|
|
210 |
do
|
211 |
export DATASET_ID=${i}
|
212 |
+
export DIRECTION="enA-frA"
|
213 |
python attach_speaker_embedding_s2s.py
|
214 |
done
|
215 |
+
for i in $(seq 381 600);
|
216 |
+
do
|
217 |
+
export DATASET_ID=${i}
|
218 |
+
export DIRECTION="enA-frA"
|
219 |
+
python attach_speaker_embedding_s2s.py
|
220 |
+
done
|
221 |
+
#TODO2
|
222 |
+
export SE_MODEL="hubert-xl"
|
223 |
+
for i in 318 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361;
|
224 |
+
do
|
225 |
+
export DATASET_ID=${i}
|
226 |
+
export DIRECTION="enA-esA"
|
227 |
+
python attach_speaker_embedding_s2s.py
|
228 |
+
done
|
229 |
+
#TODO3
|
230 |
export SE_MODEL="w2vbert-600m"
|
231 |
+
for i in $(seq 40 600);
|
232 |
+
do
|
233 |
+
export DATASET_ID=${i}
|
234 |
+
export DIRECTION="enA-frA"
|
235 |
+
python attach_speaker_embedding_s2s.py
|
236 |
+
done
|
237 |
+
#TODO4
|
238 |
+
export SE_MODEL="w2vbert-600m"
|
239 |
+
for i in $(seq 467 470);
|
240 |
+
do
|
241 |
+
export DATASET_ID=${i}
|
242 |
+
export DIRECTION="enA-esA"
|
243 |
+
python attach_speaker_embedding_s2s.py
|
244 |
+
done
|
245 |
+
export SE_MODEL="xlsr-2b"
|
246 |
+
for i in $(seq 453 470);
|
247 |
+
do
|
248 |
+
export DATASET_ID=${i}
|
249 |
+
export DIRECTION="enA-esA"
|
250 |
+
python attach_speaker_embedding_s2s.py
|
251 |
+
done
|
252 |
+
export SE_MODEL="metavoice"
|
253 |
+
for i in $(seq 453 459);
|
254 |
+
do
|
255 |
+
export DATASET_ID=${i}
|
256 |
+
export DIRECTION="enA-esA"
|
257 |
+
python attach_speaker_embedding_s2s.py
|
258 |
+
done
|
259 |
+
|
260 |
+
|
261 |
+
# running on kotoba_1
|
262 |
+
#TODO1
|
263 |
+
export SE_MODEL="hubert-xl"
|
264 |
+
for i in $(seq 1 394);
|
265 |
+
do
|
266 |
+
export DATASET_ID=${i}
|
267 |
+
export DIRECTION="deA-enA"
|
268 |
+
python attach_speaker_embedding_s2s.py
|
269 |
+
done
|
270 |
+
#TODO2
|
271 |
+
export SE_MODEL="hubert-xl"
|
272 |
+
for i in $(seq 38 600);
|
273 |
+
do
|
274 |
+
export DATASET_ID=${i}
|
275 |
+
export DIRECTION="enA-frA"
|
276 |
+
python attach_speaker_embedding_s2s.py
|
277 |
+
done
|
278 |
+
#TODO3
|
279 |
+
export SE_MODEL="metavoice"
|
280 |
+
for i in $(seq 50 299);
|
281 |
+
do
|
282 |
+
export DATASET_ID=${i}
|
283 |
+
export DIRECTION="enA-frA"
|
284 |
+
python attach_speaker_embedding_s2s.py
|
285 |
+
done
|
286 |
+
for i in $(seq 331 398);
|
287 |
+
do
|
288 |
+
export DATASET_ID=${i}
|
289 |
+
export DIRECTION="enA-frA"
|
290 |
+
python attach_speaker_embedding_s2s.py
|
291 |
+
done
|
292 |
+
for i in $(seq 426 600);
|
293 |
+
do
|
294 |
+
export DATASET_ID=${i}
|
295 |
+
export DIRECTION="enA-frA"
|
296 |
+
python attach_speaker_embedding_s2s.py
|
297 |
+
done
|
298 |
+
#TODO4
|
299 |
+
export SE_MODEL="xlsr-2b"
|
300 |
+
export DATASET_ID=213
|
301 |
+
export DIRECTION="deA-enA"
|
302 |
+
python attach_speaker_embedding_s2s.py
|
303 |
+
export SE_MODEL="hubert-xl"
|
304 |
+
for i in 117 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284;
|
305 |
do
|
306 |
export DATASET_ID=${i}
|
307 |
export DIRECTION="deA-enA"
|
|
|
313 |
|
314 |
|
315 |
|
316 |
+
|
317 |
+
|
318 |
+
|
319 |
+
|
320 |
+
|
321 |
+
|
322 |
####################
|
323 |
# enA-koA: 511_358 #
|
324 |
####################
|