add model with RIR augmentation

README.md

@@ -6,4 +6,21 @@ language:
 - en
 metrics:
 - wer
----
+---
+# LibriSpeech pruned_transducer_stateless7_streaming
+
+This model is based on the icefall `pruned_transducer_stateless7_streaming` recipe,
+but the model parameters are modified to be smaller in size. It can be
+considered a streaming version of [this model](https://huggingface.co/Zengwei/icefall-asr-librispeech-pruned-transducer-stateless7-20M-2023-01-28) and follows
+the same parameter configuration.
+
+The main difference from <https://huggingface.co/desh2608/icefall-asr-librispeech-pruned-transducer-stateless7-streaming-small> is that
+this model additionally uses simulated RIRs for training, which effectively doubles the training data.
+
+## Performance Record
+
+| Decoding method           | test-clean | test-other | 
+|---------------------------|------------|------------|
+| greedy search             |    3.58    |   9.29     |
+| fast beam search          |    3.57    |   9.05     |
+| modified beam search      |    3.41    |   8.94     |

data/lang_bpe_500/bpe.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c53433de083c4a6ad12d034550ef22de68cec62c4f58932a7b6b8b2f1e743fa5
+size 244865

data/lang_bpe_500/tokens.txt

@@ -0,0 +1,502 @@
+<blk> 0
+<sos/eos> 1
+<unk> 2
+S 3
+▁THE 4
+▁A 5
+T 6
+▁AND 7
+ED 8
+▁OF 9
+▁TO 10
+E 11
+D 12
+N 13
+ING 14
+▁IN 15
+Y 16
+M 17
+C 18
+▁I 19
+A 20
+P 21
+▁HE 22
+R 23
+O 24
+L 25
+RE 26
+I 27
+U 28
+ER 29
+▁IT 30
+LY 31
+▁THAT 32
+▁WAS 33
+▁ 34
+▁S 35
+AR 36
+▁BE 37
+F 38
+▁C 39
+IN 40
+B 41
+▁FOR 42
+OR 43
+LE 44
+' 45
+▁HIS 46
+▁YOU 47
+AL 48
+▁RE 49
+V 50
+▁B 51
+G 52
+RI 53
+▁E 54
+▁WITH 55
+▁T 56
+▁AS 57
+LL 58
+▁P 59
+▁HER 60
+ST 61
+▁HAD 62
+▁SO 63
+▁F 64
+W 65
+CE 66
+▁IS 67
+ND 68
+▁NOT 69
+TH 70
+▁BUT 71
+EN 72
+▁SHE 73
+▁ON 74
+VE 75
+ON 76
+SE 77
+▁DE 78
+UR 79
+▁G 80
+CH 81
+K 82
+TER 83
+▁AT 84
+IT 85
+▁ME 86
+RO 87
+NE 88
+RA 89
+ES 90
+IL 91
+NG 92
+IC 93
+▁NO 94
+▁HIM 95
+ENT 96
+IR 97
+▁WE 98
+H 99
+▁DO 100
+▁ALL 101
+▁HAVE 102
+LO 103
+▁BY 104
+▁MY 105
+▁MO 106
+▁THIS 107
+LA 108
+▁ST 109
+▁WHICH 110
+▁CON 111
+▁THEY 112
+CK 113
+TE 114
+▁SAID 115
+▁FROM 116
+▁GO 117
+▁WHO 118
+▁TH 119
+▁OR 120
+▁D 121
+▁W 122
+VER 123
+LI 124
+▁SE 125
+▁ONE 126
+▁CA 127
+▁AN 128
+▁LA 129
+▁WERE 130
+EL 131
+▁HA 132
+▁MAN 133
+▁FA 134
+▁EX 135
+AD 136
+▁SU 137
+RY 138
+▁MI 139
+AT 140
+▁BO 141
+▁WHEN 142
+AN 143
+THER 144
+PP 145
+ATION 146
+▁FI 147
+▁WOULD 148
+▁PRO 149
+OW 150
+ET 151
+▁O 152
+▁THERE 153
+▁HO 154
+ION 155
+▁WHAT 156
+▁FE 157
+▁PA 158
+US 159
+MENT 160
+▁MA 161
+UT 162
+▁OUT 163
+▁THEIR 164
+▁IF 165
+▁LI 166
+▁K 167
+▁WILL 168
+▁ARE 169
+ID 170
+▁RO 171
+DE 172
+TION 173
+▁WA 174
+PE 175
+▁UP 176
+▁SP 177
+▁PO 178
+IGHT 179
+▁UN 180
+RU 181
+▁LO 182
+AS 183
+OL 184
+▁LE 185
+▁BEEN 186
+▁SH 187
+▁RA 188
+▁SEE 189
+KE 190
+UL 191
+TED 192
+▁SA 193
+UN 194
+UND 195
+ANT 196
+▁NE 197
+IS 198
+▁THEM 199
+CI 200
+GE 201
+▁COULD 202
+▁DIS 203
+OM 204
+ISH 205
+HE 206
+EST 207
+▁SOME 208
+ENCE 209
+ITY 210
+IVE 211
+▁US 212
+▁MORE 213
+▁EN 214
+ARD 215
+ATE 216
+▁YOUR 217
+▁INTO 218
+▁KNOW 219
+▁CO 220
+ANCE 221
+▁TIME 222
+▁WI 223
+▁YE 224
+AGE 225
+▁NOW 226
+TI 227
+FF 228
+ABLE 229
+▁VERY 230
+▁LIKE 231
+AM 232
+HI 233
+Z 234
+▁OTHER 235
+▁THAN 236
+▁LITTLE 237
+▁DID 238
+▁LOOK 239
+TY 240
+ERS 241
+▁CAN 242
+▁CHA 243
+▁AR 244
+X 245
+FUL 246
+UGH 247
+▁BA 248
+▁DAY 249
+▁ABOUT 250
+TEN 251
+IM 252
+▁ANY 253
+▁PRE 254
+▁OVER 255
+IES 256
+NESS 257
+ME 258
+BLE 259
+▁M 260
+ROW 261
+▁HAS 262
+▁GREAT 263
+▁VI 264
+TA 265
+▁AFTER 266
+PER 267
+▁AGAIN 268
+HO 269
+SH 270
+▁UPON 271
+▁DI 272
+▁HAND 273
+▁COM 274
+IST 275
+TURE 276
+▁STA 277
+▁THEN 278
+▁SHOULD 279
+▁GA 280
+OUS 281
+OUR 282
+▁WELL 283
+▁ONLY 284
+MAN 285
+▁GOOD 286
+▁TWO 287
+▁MAR 288
+▁SAY 289
+▁HU 290
+TING 291
+▁OUR 292
+RESS 293
+▁DOWN 294
+IOUS 295
+▁BEFORE 296
+▁DA 297
+▁NA 298
+QUI 299
+▁MADE 300
+▁EVERY 301
+▁OLD 302
+▁EVEN 303
+IG 304
+▁COME 305
+▁GRA 306
+▁RI 307
+▁LONG 308
+OT 309
+SIDE 310
+WARD 311
+▁FO 312
+▁WHERE 313
+MO 314
+LESS 315
+▁SC 316
+▁MUST 317
+▁NEVER 318
+▁HOW 319
+▁CAME 320
+▁SUCH 321
+▁RU 322
+▁TAKE 323
+▁WO 324
+▁CAR 325
+UM 326
+AK 327
+▁THINK 328
+▁MUCH 329
+▁MISTER 330
+▁MAY 331
+▁JO 332
+▁WAY 333
+▁COMP 334
+▁THOUGHT 335
+▁STO 336
+▁MEN 337
+▁BACK 338
+▁DON 339
+J 340
+▁LET 341
+▁TRA 342
+▁FIRST 343
+▁JUST 344
+▁VA 345
+▁OWN 346
+▁PLA 347
+▁MAKE 348
+ATED 349
+▁HIMSELF 350
+▁WENT 351
+▁PI 352
+GG 353
+RING 354
+▁DU 355
+▁MIGHT 356
+▁PART 357
+▁GIVE 358
+▁IMP 359
+▁BU 360
+▁PER 361
+▁PLACE 362
+▁HOUSE 363
+▁THROUGH 364
+IAN 365
+▁SW 366
+▁UNDER 367
+QUE 368
+▁AWAY 369
+▁LOVE 370
+QUA 371
+▁LIFE 372
+▁GET 373
+▁WITHOUT 374
+▁PASS 375
+▁TURN 376
+IGN 377
+▁HEAD 378
+▁MOST 379
+▁THOSE 380
+▁SHALL 381
+▁EYES 382
+▁COL 383
+▁STILL 384
+▁NIGHT 385
+▁NOTHING 386
+ITION 387
+HA 388
+▁TELL 389
+▁WORK 390
+▁LAST 391
+▁NEW 392
+▁FACE 393
+▁HI 394
+▁WORD 395
+▁FOUND 396
+▁COUNT 397
+▁OB 398
+▁WHILE 399
+▁SHA 400
+▁MEAN 401
+▁SAW 402
+▁PEOPLE 403
+▁FRIEND 404
+▁THREE 405
+▁ROOM 406
+▁SAME 407
+▁THOUGH 408
+▁RIGHT 409
+▁CHILD 410
+▁FATHER 411
+▁ANOTHER 412
+▁HEART 413
+▁WANT 414
+▁TOOK 415
+OOK 416
+▁LIGHT 417
+▁MISSUS 418
+▁OPEN 419
+▁JU 420
+▁ASKED 421
+PORT 422
+▁LEFT 423
+▁JA 424
+▁WORLD 425
+▁HOME 426
+▁WHY 427
+▁ALWAYS 428
+▁ANSWER 429
+▁SEEMED 430
+▁SOMETHING 431
+▁GIRL 432
+▁BECAUSE 433
+▁NAME 434
+▁TOLD 435
+▁NI 436
+▁HIGH 437
+IZE 438
+▁WOMAN 439
+▁FOLLOW 440
+▁RETURN 441
+▁KNEW 442
+▁EACH 443
+▁KIND 444
+▁JE 445
+▁ACT 446
+▁LU 447
+▁CERTAIN 448
+▁YEARS 449
+▁QUITE 450
+▁APPEAR 451
+▁BETTER 452
+▁HALF 453
+▁PRESENT 454
+▁PRINCE 455
+SHIP 456
+▁ALSO 457
+▁BEGAN 458
+▁HAVING 459
+▁ENOUGH 460
+▁PERSON 461
+▁LADY 462
+▁WHITE 463
+▁COURSE 464
+▁VOICE 465
+▁SPEAK 466
+▁POWER 467
+▁MORNING 468
+▁BETWEEN 469
+▁AMONG 470
+▁KEEP 471
+▁WALK 472
+▁MATTER 473
+▁TEA 474
+▁BELIEVE 475
+▁SMALL 476
+▁TALK 477
+▁FELT 478
+▁HORSE 479
+▁MYSELF 480
+▁SIX 481
+▁HOWEVER 482
+▁FULL 483
+▁HERSELF 484
+▁POINT 485
+▁STOOD 486
+▁HUNDRED 487
+▁ALMOST 488
+▁SINCE 489
+▁LARGE 490
+▁LEAVE 491
+▁PERHAPS 492
+▁DARK 493
+▁SUDDEN 494
+▁REPLIED 495
+▁ANYTHING 496
+▁WONDER 497
+▁UNTIL 498
+Q 499
+#0 500
+#1 501

exp/cpu_jit.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:590f80b5b73e8b207121113e18ea6cb0f4254e415a9af6c19ab43b7611e4cd24
+size 134187228

exp/decoder_jit_trace.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23520c0ed0d6738d9dd52584296394b9b9ec577c930b6150404e4508f3ac8381
+size 1047141

exp/encoder_jit_trace.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c67b2fef64df0fdefebd8a45cd45f16c5f43d6b2a156ef70cb7cd40654324cd5
+size 129943703

exp/export.sh

@@ -0,0 +1,18 @@
+./pruned_transducer_stateless7_streaming/export.py \
+  --bpe-model data/lang_bpe_500/bpe.model \
+  --use-averaged-model=True \
+  --epoch 30 \
+  --avg 9 \
+  --decode-chunk-len 32 \
+  --jit 0 \
+  --exp-dir ./pruned_transducer_stateless7_streaming/exp/ \
+  --num-encoder-layers 2,2,2,2,2 \
+  --feedforward-dims 768,768,768,768,768 \
+  --nhead 8,8,8,8,8 \
+  --encoder-dims 256,256,256,256,256 \
+  --attention-dims 192,192,192,192,192 \
+  --encoder-unmasked-dims 192,192,192,192,192 \
+  --zipformer-downsampling-factors 1,2,4,8,2 \
+  --cnn-module-kernels 31,31,31,31,31 \
+  --decoder-dim 512 \
+  --joiner-dim 512

exp/jit_trace_export.sh

@@ -0,0 +1,17 @@
+./pruned_transducer_stateless7_streaming/jit_trace_export.py \
+  --bpe-model data/lang_bpe_500/bpe.model \
+  --use-averaged-model=True \
+  --epoch 30 \
+  --avg 9 \
+  --decode-chunk-len 32 \
+  --exp-dir ./pruned_transducer_stateless7_streaming/exp \
+  --num-encoder-layers 2,2,2,2,2 \
+  --feedforward-dims 768,768,768,768,768 \
+  --nhead 8,8,8,8,8 \
+  --encoder-dims 256,256,256,256,256 \
+  --attention-dims 192,192,192,192,192 \
+  --encoder-unmasked-dims 192,192,192,192,192 \
+  --zipformer-downsampling-factors 1,2,4,8,2 \
+  --cnn-module-kernels 31,31,31,31,31 \
+  --decoder-dim 512 \
+  --joiner-dim 512

exp/joiner_jit_trace.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ecb6e215a43492b243ad4af3adf2cf5b761987fdb5a92656453b0e7b16c6de6
+size 2611547

exp/pretrained.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfffe9f495d6a0599cc77c6b415ae04a18baa0446b3d73b7a431867296fd0de1
+size 82988698

exp/tensorboard/events.out.tfevents.1680303114.r2n03.443763.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abfff0db01df52f1470a881db15a291e852a7c84aa0c26a25c98acc7542b5773
+size 1984002

log/fast_beam_search/log-decode-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model-2023-04-04-09-21-03

@@ -0,0 +1,45 @@
+2023-04-04 09:21:03,151 INFO [decode.py:649] Decoding started
+2023-04-04 09:21:03,151 INFO [decode.py:655] Device: cuda:0
+2023-04-04 09:21:03,214 INFO [decode.py:665] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.3', 'k2-build-type': 'Debug', 'k2-with-cuda': True, 'k2-git-sha1': '1c9950559223ec24d187f56bc424c3b43904bed3', 'k2-git-date': 'Thu Jan 26 22:00:26 2023', 'lhotse-version': '1.13.0.dev+git.ca98c73.dirty', 'torch-version': '2.0.0+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.8', 'icefall-git-branch': 'surt', 'icefall-git-sha1': '51e6a8a-dirty', 'icefall-git-date': 'Fri Mar 17 11:23:13 2023', 'icefall-path': '/exp/draj/mini_scale_2022/icefall', 'k2-path': '/exp/draj/mini_scale_2022/k2/k2/python/k2/__init__.py', 'lhotse-path': '/exp/draj/mini_scale_2022/lhotse/lhotse/__init__.py', 'hostname': 'r7n04', 'IP address': '10.1.7.4'}, 'epoch': 30, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming/exp/v2'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'fast_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 4, 'max_states': 8, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'num_encoder_layers': '2,2,2,2,2', 'feedforward_dims': '768,768,768,768,768', 'nhead': '8,8,8,8,8', 'encoder_dims': '256,256,256,256,256', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '192,192,192,192,192', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'full_libri': True, 'manifest_dir': PosixPath('data/manifests'), 'max_duration': 500, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'res_dir': PosixPath('pruned_transducer_stateless7_streaming/exp/v2/fast_beam_search'), 'suffix': 'epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 09:21:03,214 INFO [decode.py:667] About to create model
+2023-04-04 09:21:03,641 INFO [zipformer.py:405] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2023-04-04 09:21:03,649 INFO [decode.py:738] Calculating the averaged model over epoch range from 21 (excluded) to 30
+2023-04-04 09:21:12,177 INFO [decode.py:772] Number of model parameters: 20697573
+2023-04-04 09:21:12,178 INFO [asr_datamodule.py:454] About to get test-clean cuts
+2023-04-04 09:21:12,204 INFO [asr_datamodule.py:461] About to get test-other cuts
+2023-04-04 09:21:21,894 INFO [decode.py:560] batch 0/?, cuts processed until now is 36
+2023-04-04 09:22:03,674 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.3765, 1.4193, 1.7228, 1.7642, 1.3093, 1.6687, 1.6459, 1.5270],
+       device='cuda:0'), covar=tensor([0.3476, 0.3824, 0.1703, 0.2398, 0.3968, 0.2215, 0.4112, 0.3021],
+       device='cuda:0'), in_proj_covar=tensor([0.0922, 0.0996, 0.0730, 0.0941, 0.0899, 0.0836, 0.0850, 0.0796],
+       device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002],
+       device='cuda:0')
+2023-04-04 09:22:18,478 INFO [decode.py:560] batch 20/?, cuts processed until now is 1038
+2023-04-04 09:23:05,390 INFO [decode.py:560] batch 40/?, cuts processed until now is 2296
+2023-04-04 09:23:30,148 INFO [decode.py:574] The transcripts are stored in pruned_transducer_stateless7_streaming/exp/v2/fast_beam_search/recogs-test-clean-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model.txt
+2023-04-04 09:23:30,221 INFO [utils.py:560] [test-clean-beam_20.0_max_contexts_4_max_states_8] %WER 3.57% [1879 / 52576, 218 ins, 142 del, 1519 sub ]
+2023-04-04 09:23:30,377 INFO [decode.py:585] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp/v2/fast_beam_search/errs-test-clean-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model.txt
+2023-04-04 09:23:30,378 INFO [decode.py:599] 
+For test-clean, WER of different settings are:
+beam_20.0_max_contexts_4_max_states_8	3.57	best for test-clean
+
+2023-04-04 09:23:33,849 INFO [decode.py:560] batch 0/?, cuts processed until now is 43
+2023-04-04 09:24:24,002 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.2127, 1.4455, 1.7780, 1.1393, 2.3796, 2.9278, 2.6542, 2.9877],
+       device='cuda:0'), covar=tensor([0.1591, 0.3740, 0.3306, 0.2715, 0.0603, 0.0200, 0.0263, 0.0375],
+       device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0327, 0.0358, 0.0267, 0.0247, 0.0189, 0.0215, 0.0266],
+       device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003],
+       device='cuda:0')
+2023-04-04 09:24:25,372 INFO [decode.py:560] batch 20/?, cuts processed until now is 1198
+2023-04-04 09:24:38,009 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.5290, 1.4286, 1.4607, 1.8452, 1.4408, 1.7147, 1.6404, 1.6043],
+       device='cuda:0'), covar=tensor([0.0797, 0.0904, 0.0939, 0.0609, 0.0893, 0.0737, 0.0895, 0.0659],
+       device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0220, 0.0224, 0.0236, 0.0223, 0.0210, 0.0185, 0.0202],
+       device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004],
+       device='cuda:0')
+2023-04-04 09:25:11,694 INFO [decode.py:560] batch 40/?, cuts processed until now is 2642
+2023-04-04 09:25:33,579 INFO [decode.py:574] The transcripts are stored in pruned_transducer_stateless7_streaming/exp/v2/fast_beam_search/recogs-test-other-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model.txt
+2023-04-04 09:25:33,661 INFO [utils.py:560] [test-other-beam_20.0_max_contexts_4_max_states_8] %WER 9.05% [4738 / 52343, 515 ins, 457 del, 3766 sub ]
+2023-04-04 09:25:33,838 INFO [decode.py:585] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp/v2/fast_beam_search/errs-test-other-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model.txt
+2023-04-04 09:25:33,839 INFO [decode.py:599] 
+For test-other, WER of different settings are:
+beam_20.0_max_contexts_4_max_states_8	9.05	best for test-other
+
+2023-04-04 09:25:33,839 INFO [decode.py:803] Done!

log/fast_beam_search/wer-summary-test-clean-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_4_max_states_8	3.57

log/fast_beam_search/wer-summary-test-other-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-4-max-states-8-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_20.0_max_contexts_4_max_states_8	9.05

log/greedy_search/log-decode-epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model-2023-04-04-09-35-47

@@ -0,0 +1,33 @@
+2023-04-04 09:35:47,107 INFO [decode.py:649] Decoding started
+2023-04-04 09:35:47,108 INFO [decode.py:655] Device: cuda:0
+2023-04-04 09:35:47,110 INFO [decode.py:665] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.3', 'k2-build-type': 'Debug', 'k2-with-cuda': True, 'k2-git-sha1': '1c9950559223ec24d187f56bc424c3b43904bed3', 'k2-git-date': 'Thu Jan 26 22:00:26 2023', 'lhotse-version': '1.13.0.dev+git.ca98c73.dirty', 'torch-version': '2.0.0+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.8', 'icefall-git-branch': 'surt', 'icefall-git-sha1': '51e6a8a-dirty', 'icefall-git-date': 'Fri Mar 17 11:23:13 2023', 'icefall-path': '/exp/draj/mini_scale_2022/icefall', 'k2-path': '/exp/draj/mini_scale_2022/k2/k2/python/k2/__init__.py', 'lhotse-path': '/exp/draj/mini_scale_2022/lhotse/lhotse/__init__.py', 'hostname': 'r7n04', 'IP address': '10.1.7.4'}, 'epoch': 30, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming/exp/v2'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 4, 'max_states': 8, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'num_encoder_layers': '2,2,2,2,2', 'feedforward_dims': '768,768,768,768,768', 'nhead': '8,8,8,8,8', 'encoder_dims': '256,256,256,256,256', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '192,192,192,192,192', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'full_libri': True, 'manifest_dir': PosixPath('data/manifests'), 'max_duration': 500, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'res_dir': PosixPath('pruned_transducer_stateless7_streaming/exp/v2/greedy_search'), 'suffix': 'epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 09:35:47,110 INFO [decode.py:667] About to create model
+2023-04-04 09:35:47,453 INFO [zipformer.py:405] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2023-04-04 09:35:47,461 INFO [decode.py:738] Calculating the averaged model over epoch range from 21 (excluded) to 30
+2023-04-04 09:35:49,999 INFO [decode.py:772] Number of model parameters: 20697573
+2023-04-04 09:35:49,999 INFO [asr_datamodule.py:454] About to get test-clean cuts
+2023-04-04 09:35:50,001 INFO [asr_datamodule.py:461] About to get test-other cuts
+2023-04-04 09:35:53,538 INFO [decode.py:560] batch 0/?, cuts processed until now is 36
+2023-04-04 09:36:38,926 INFO [decode.py:560] batch 50/?, cuts processed until now is 2609
+2023-04-04 09:36:40,063 INFO [decode.py:574] The transcripts are stored in pruned_transducer_stateless7_streaming/exp/v2/greedy_search/recogs-test-clean-epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-04 09:36:40,141 INFO [utils.py:560] [test-clean-greedy_search] %WER 3.58% [1881 / 52576, 211 ins, 163 del, 1507 sub ]
+2023-04-04 09:36:40,301 INFO [decode.py:585] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp/v2/greedy_search/errs-test-clean-epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-04 09:36:40,302 INFO [decode.py:599] 
+For test-clean, WER of different settings are:
+greedy_search	3.58	best for test-clean
+
+2023-04-04 09:36:41,891 INFO [decode.py:560] batch 0/?, cuts processed until now is 43
+2023-04-04 09:36:45,297 INFO [zipformer.py:2441] attn_weights_entropy = tensor([3.1587, 1.3790, 1.6006, 1.5477, 2.7838, 1.2051, 2.2557, 3.1271],
+       device='cuda:0'), covar=tensor([0.0570, 0.2876, 0.2844, 0.1718, 0.0661, 0.2407, 0.1209, 0.0258],
+       device='cuda:0'), in_proj_covar=tensor([0.0415, 0.0371, 0.0391, 0.0347, 0.0375, 0.0351, 0.0386, 0.0408],
+       device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003],
+       device='cuda:0')
+2023-04-04 09:37:23,322 INFO [decode.py:560] batch 50/?, cuts processed until now is 2939
+2023-04-04 09:37:23,429 INFO [decode.py:574] The transcripts are stored in pruned_transducer_stateless7_streaming/exp/v2/greedy_search/recogs-test-other-epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-04 09:37:23,509 INFO [utils.py:560] [test-other-greedy_search] %WER 9.29% [4862 / 52343, 500 ins, 491 del, 3871 sub ]
+2023-04-04 09:37:23,678 INFO [decode.py:585] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp/v2/greedy_search/errs-test-other-epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt
+2023-04-04 09:37:23,679 INFO [decode.py:599] 
+For test-other, WER of different settings are:
+greedy_search	9.29	best for test-other
+
+2023-04-04 09:37:23,679 INFO [decode.py:803] Done!

log/greedy_search/wer-summary-test-clean-epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+greedy_search	3.58

log/greedy_search/wer-summary-test-other-epoch-30-avg-9-streaming-chunk-size-32-context-2-max-sym-per-frame-1-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+greedy_search	9.29

log/modified_beam_search/log-decode-epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model-2023-04-04-09-26-31

@@ -0,0 +1,35 @@
+2023-04-04 09:26:31,976 INFO [decode.py:649] Decoding started
+2023-04-04 09:26:31,976 INFO [decode.py:655] Device: cuda:0
+2023-04-04 09:26:31,978 INFO [decode.py:665] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.3', 'k2-build-type': 'Debug', 'k2-with-cuda': True, 'k2-git-sha1': '1c9950559223ec24d187f56bc424c3b43904bed3', 'k2-git-date': 'Thu Jan 26 22:00:26 2023', 'lhotse-version': '1.13.0.dev+git.ca98c73.dirty', 'torch-version': '2.0.0+cu117', 'torch-cuda-available': True, 'torch-cuda-version': '11.7', 'python-version': '3.8', 'icefall-git-branch': 'surt', 'icefall-git-sha1': '51e6a8a-dirty', 'icefall-git-date': 'Fri Mar 17 11:23:13 2023', 'icefall-path': '/exp/draj/mini_scale_2022/icefall', 'k2-path': '/exp/draj/mini_scale_2022/k2/k2/python/k2/__init__.py', 'lhotse-path': '/exp/draj/mini_scale_2022/lhotse/lhotse/__init__.py', 'hostname': 'r7n04', 'IP address': '10.1.7.4'}, 'epoch': 30, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming/exp/v2'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'modified_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 4, 'max_states': 8, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'num_encoder_layers': '2,2,2,2,2', 'feedforward_dims': '768,768,768,768,768', 'nhead': '8,8,8,8,8', 'encoder_dims': '256,256,256,256,256', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '192,192,192,192,192', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'full_libri': True, 'manifest_dir': PosixPath('data/manifests'), 'max_duration': 500, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures', 'res_dir': PosixPath('pruned_transducer_stateless7_streaming/exp/v2/modified_beam_search'), 'suffix': 'epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
+2023-04-04 09:26:31,979 INFO [decode.py:667] About to create model
+2023-04-04 09:26:32,322 INFO [zipformer.py:405] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
+2023-04-04 09:26:32,330 INFO [decode.py:738] Calculating the averaged model over epoch range from 21 (excluded) to 30
+2023-04-04 09:26:34,928 INFO [decode.py:772] Number of model parameters: 20697573
+2023-04-04 09:26:34,928 INFO [asr_datamodule.py:454] About to get test-clean cuts
+2023-04-04 09:26:34,931 INFO [asr_datamodule.py:461] About to get test-other cuts
+2023-04-04 09:26:42,421 INFO [decode.py:560] batch 0/?, cuts processed until now is 36
+2023-04-04 09:28:20,169 INFO [decode.py:560] batch 20/?, cuts processed until now is 1038
+2023-04-04 09:29:48,445 INFO [decode.py:560] batch 40/?, cuts processed until now is 2296
+2023-04-04 09:30:18,365 INFO [decode.py:574] The transcripts are stored in pruned_transducer_stateless7_streaming/exp/v2/modified_beam_search/recogs-test-clean-epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
+2023-04-04 09:30:18,575 INFO [utils.py:560] [test-clean-beam_size_4] %WER 3.41% [1792 / 52576, 213 ins, 129 del, 1450 sub ]
+2023-04-04 09:30:18,740 INFO [decode.py:585] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp/v2/modified_beam_search/errs-test-clean-epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
+2023-04-04 09:30:18,741 INFO [decode.py:599] 
+For test-clean, WER of different settings are:
+beam_size_4	3.41	best for test-clean
+
+2023-04-04 09:30:24,057 INFO [decode.py:560] batch 0/?, cuts processed until now is 43
+2023-04-04 09:31:28,835 INFO [zipformer.py:2441] attn_weights_entropy = tensor([1.5981, 1.5829, 1.6027, 1.3932, 1.3421, 1.3843, 0.4039, 0.7473],
+       device='cuda:0'), covar=tensor([0.0687, 0.0691, 0.0417, 0.0690, 0.1242, 0.0840, 0.1421, 0.1122],
+       device='cuda:0'), in_proj_covar=tensor([0.0357, 0.0356, 0.0360, 0.0384, 0.0463, 0.0389, 0.0338, 0.0341],
+       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002],
+       device='cuda:0')
+2023-04-04 09:31:56,538 INFO [decode.py:560] batch 20/?, cuts processed until now is 1198
+2023-04-04 09:33:26,353 INFO [decode.py:560] batch 40/?, cuts processed until now is 2642
+2023-04-04 09:33:50,079 INFO [decode.py:574] The transcripts are stored in pruned_transducer_stateless7_streaming/exp/v2/modified_beam_search/recogs-test-other-epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
+2023-04-04 09:33:50,162 INFO [utils.py:560] [test-other-beam_size_4] %WER 8.94% [4681 / 52343, 512 ins, 424 del, 3745 sub ]
+2023-04-04 09:33:50,334 INFO [decode.py:585] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp/v2/modified_beam_search/errs-test-other-epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
+2023-04-04 09:33:50,334 INFO [decode.py:599] 
+For test-other, WER of different settings are:
+beam_size_4	8.94	best for test-other
+
+2023-04-04 09:33:50,335 INFO [decode.py:803] Done!

log/modified_beam_search/wer-summary-test-clean-epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_size_4	3.41

log/modified_beam_search/wer-summary-test-other-epoch-30-avg-9-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt

@@ -0,0 +1,2 @@
+settings	WER
+beam_size_4	8.94

test_wavs/trans.txt

@@ -0,0 +1,3 @@
+1089-134686-0001 AFTER EARLY NIGHTFALL THE YELLOW LAMPS WOULD LIGHT UP HERE AND THERE THE SQUALID QUARTER OF THE BROTHELS
+1221-135766-0001 GOD AS A DIRECT CONSEQUENCE OF THE SIN WHICH MAN THUS PUNISHED HAD GIVEN HER A LOVELY CHILD WHOSE PLACE WAS ON THAT SAME DISHONOURED BOSOM TO CONNECT HER PARENT FOR EVER WITH THE RACE AND DESCENT OF MORTALS AND TO BE FINALLY A BLESSED SOUL IN HEAVEN
+1221-135766-0002 YET THESE THOUGHTS AFFECTED HESTER PRYNNE LESS WITH HOPE THAN APPREHENSION