forked from alibaba/Alibaba-MIT-Speech
-
Notifications
You must be signed in to change notification settings - Fork 0
/
RESULTS
797 lines (712 loc) · 79.9 KB
/
RESULTS
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
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
# In the results below, "tgsmall" is the pruned 3-gram LM, which is used for lattice generation.
# The following language models are then used for rescoring:
# a) tgmed- slightly less pruned 3-gram LM
# b) tglarge- the full, non-pruned 3-gram LM
# c) fglarge- non-pruned 4-gram LM
#
# The "dev-clean" and "test-clean" sets generally contain, relatively cleaner US English acccented speech,
# whereas "dev-other" and "test-other" sets contain more challenging speech
### SAT GMM model trained on the "train-clean-100" set (100 hours "clean" speech)
### for test in dev_clean test_clean dev_other test_other; do for lm in fglarge tglarge tgmed tgsmall; do grep WER exp/tri4b/decode_${lm}_${test}/wer* | best_wer.sh; done; echo; done
%WER 8.20 [ 4459 / 54402, 695 ins, 427 del, 3337 sub ] exp/tri4b/decode_fglarge_dev_clean/wer_14_0.5
%WER 8.60 [ 4677 / 54402, 763 ins, 399 del, 3515 sub ] exp/tri4b/decode_tglarge_dev_clean/wer_16_0.0
%WER 10.39 [ 5655 / 54402, 711 ins, 648 del, 4296 sub ] exp/tri4b/decode_tgmed_dev_clean/wer_16_0.0
%WER 11.69 [ 6361 / 54402, 743 ins, 808 del, 4810 sub ] exp/tri4b/decode_tgsmall_dev_clean/wer_16_0.0
%WER 9.10 [ 4786 / 52576, 708 ins, 464 del, 3614 sub ] exp/tri4b/decode_fglarge_test_clean/wer_17_0.5
%WER 9.43 [ 4958 / 52576, 751 ins, 492 del, 3715 sub ] exp/tri4b/decode_tglarge_test_clean/wer_15_0.5
%WER 11.36 [ 5975 / 52576, 799 ins, 642 del, 4534 sub ] exp/tri4b/decode_tgmed_test_clean/wer_17_0.0
%WER 12.64 [ 6643 / 52576, 795 ins, 817 del, 5031 sub ] exp/tri4b/decode_tgsmall_test_clean/wer_17_0.0
%WER 28.45 [ 14495 / 50948, 1574 ins, 1925 del, 10996 sub ] exp/tri4b/decode_fglarge_dev_other/wer_17_0.5
%WER 29.24 [ 14895 / 50948, 1610 ins, 2041 del, 11244 sub ] exp/tri4b/decode_tglarge_dev_other/wer_19_0.5
%WER 32.04 [ 16325 / 50948, 1753 ins, 2261 del, 12311 sub ] exp/tri4b/decode_tgmed_dev_other/wer_18_0.0
%WER 33.97 [ 17305 / 50948, 1681 ins, 2661 del, 12963 sub ] exp/tri4b/decode_tgsmall_dev_other/wer_18_0.0
%WER 30.33 [ 15875 / 52343, 1639 ins, 2375 del, 11861 sub ] exp/tri4b/decode_fglarge_test_other/wer_17_0.5
%WER 31.07 [ 16264 / 52343, 1728 ins, 2424 del, 12112 sub ] exp/tri4b/decode_tglarge_test_other/wer_18_0.5
%WER 33.69 [ 17633 / 52343, 1755 ins, 2766 del, 13112 sub ] exp/tri4b/decode_tgmed_test_other/wer_18_0.0
%WER 35.62 [ 18646 / 52343, 1758 ins, 3039 del, 13849 sub ] exp/tri4b/decode_tgsmall_test_other/wer_17_0.0
### SAT GMM model trained on the combined "train-clean-100" + "train-clean-360" set (460 hours "clean" speech)
### for test in dev_clean test_clean dev_other test_other; do for lm in fglarge tglarge tgmed tgsmall; do grep WER exp/tri5b/decode_${lm}_${test}/wer* | best_wer.sh; done; echo; done
%WER 7.05 [ 3835 / 54402, 588 ins, 370 del, 2877 sub ] exp/tri5b/decode_fglarge_dev_clean/wer_15_0.5
%WER 7.49 [ 4077 / 54402, 623 ins, 376 del, 3078 sub ] exp/tri5b/decode_tglarge_dev_clean/wer_14_0.5
%WER 9.38 [ 5104 / 54402, 701 ins, 533 del, 3870 sub ] exp/tri5b/decode_tgmed_dev_clean/wer_15_0.0
%WER 10.51 [ 5719 / 54402, 720 ins, 652 del, 4347 sub ] exp/tri5b/decode_tgsmall_dev_clean/wer_15_0.0
%WER 8.14 [ 4279 / 52576, 683 ins, 379 del, 3217 sub ] exp/tri5b/decode_fglarge_test_clean/wer_15_0.5
%WER 8.50 [ 4469 / 52576, 597 ins, 510 del, 3362 sub ] exp/tri5b/decode_tglarge_test_clean/wer_15_1.0
%WER 10.10 [ 5311 / 52576, 767 ins, 503 del, 4041 sub ] exp/tri5b/decode_tgmed_test_clean/wer_15_0.0
%WER 11.20 [ 5886 / 52576, 774 ins, 617 del, 4495 sub ] exp/tri5b/decode_tgsmall_test_clean/wer_15_0.0
%WER 25.65 [ 13069 / 50948, 1664 ins, 1486 del, 9919 sub ] exp/tri5b/decode_fglarge_dev_other/wer_18_0.0
%WER 26.60 [ 13552 / 50948, 1549 ins, 1774 del, 10229 sub ] exp/tri5b/decode_tglarge_dev_other/wer_17_0.5
%WER 29.21 [ 14880 / 50943, 1618 ins, 2026 del, 11236 sub ] exp/tri5b/decode_tgmed_dev_other/wer_18_0.0
%WER 30.89 [ 15736 / 50948, 1538 ins, 2388 del, 11810 sub ] exp/tri5b/decode_tgsmall_dev_other/wer_18_0.0
%WER 27.36 [ 14323 / 52343, 1486 ins, 2136 del, 10701 sub ] exp/tri5b/decode_fglarge_test_other/wer_17_0.5
%WER 28.32 [ 14824 / 52343, 1656 ins, 2118 del, 11050 sub ] exp/tri5b/decode_tglarge_test_other/wer_16_0.5
%WER 31.01 [ 16233 / 52343, 1577 ins, 2593 del, 12063 sub ] exp/tri5b/decode_tgmed_test_other/wer_19_0.0
%WER 32.99 [ 17269 / 52343, 1622 ins, 2792 del, 12855 sub ] exp/tri5b/decode_tgsmall_test_other/wer_17_0.0
### SAT GMM model trained on the combined "train-clean-100" + "train-clean-360" + "train-other-500" set (960 hours)
### for test in dev_clean test_clean dev_other test_other; do for lm in fglarge tglarge tgmed tgsmall; do grep WER exp/tri6b/decode_${lm}_${test}/wer* | best_wer.sh; done; echo; done
%WER 7.02 [ 3819 / 54402, 516 ins, 424 del, 2879 sub ] exp/tri6b/decode_fglarge_dev_clean/wer_14_1.0
%WER 7.33 [ 3988 / 54402, 506 ins, 468 del, 3014 sub ] exp/tri6b/decode_tglarge_dev_clean/wer_15_1.0
%WER 9.23 [ 5024 / 54402, 744 ins, 481 del, 3799 sub ] exp/tri6b/decode_tgmed_dev_clean/wer_13_0.0
%WER 10.38 [ 5648 / 54402, 741 ins, 617 del, 4290 sub ] exp/tri6b/decode_tgsmall_dev_clean/wer_14_0.0
%WER 7.81 [ 4105 / 52576, 574 ins, 442 del, 3089 sub ] exp/tri6b/decode_fglarge_test_clean/wer_15_1.0
%WER 8.01 [ 4213 / 52576, 658 ins, 387 del, 3168 sub ] exp/tri6b/decode_tglarge_test_clean/wer_15_0.5
%WER 9.83 [ 5167 / 52576, 709 ins, 519 del, 3939 sub ] exp/tri6b/decode_tgmed_test_clean/wer_16_0.0
%WER 10.99 [ 5778 / 52576, 723 ins, 640 del, 4415 sub ] exp/tri6b/decode_tgsmall_test_clean/wer_16_0.0
%WER 20.53 [ 10460 / 50948, 1270 ins, 1258 del, 7932 sub ] exp/tri6b/decode_fglarge_dev_other/wer_15_0.5
%WER 21.31 [ 10857 / 50948, 1299 ins, 1376 del, 8182 sub ] exp/tri6b/decode_tglarge_dev_other/wer_16_0.5
%WER 24.27 [ 12365 / 50948, 1401 ins, 1558 del, 9406 sub ] exp/tri6b/decode_tgmed_dev_other/wer_16_0.0
%WER 26.14 [ 13317 / 50948, 1292 ins, 1977 del, 10048 sub ] exp/tri6b/decode_tgsmall_dev_other/wer_17_0.0
%WER 21.79 [ 11405 / 52343, 1263 ins, 1497 del, 8645 sub ] exp/tri6b/decode_fglarge_test_other/wer_15_0.5
%WER 22.78 [ 11923 / 52343, 1370 ins, 1483 del, 9070 sub ] exp/tri6b/decode_tglarge_test_other/wer_14_0.5
%WER 25.67 [ 13439 / 52343, 1481 ins, 1767 del, 10191 sub ] exp/tri6b/decode_tgmed_test_other/wer_15_0.0
%WER 27.79 [ 14545 / 52343, 1371 ins, 2250 del, 10924 sub ] exp/tri6b/decode_tgsmall_test_other/wer_16_0.0
### p-norm DNN trained on "train-clean-100"
%WER 5.93 [ 3228 / 54402, 486 ins, 330 del, 2412 sub ] exp/nnet5a_clean_100_gpu/decode_fglarge_dev_clean/wer_13
%WER 6.32 [ 3438 / 54402, 517 ins, 365 del, 2556 sub ] exp/nnet5a_clean_100_gpu/decode_tglarge_dev_clean/wer_12
%WER 7.91 [ 4304 / 54402, 468 ins, 611 del, 3225 sub ] exp/nnet5a_clean_100_gpu/decode_tgmed_dev_clean/wer_13
%WER 9.19 [ 4998 / 54402, 567 ins, 708 del, 3723 sub ] exp/nnet5a_clean_100_gpu/decode_tgsmall_dev_clean/wer_11
%WER 6.59 [ 3464 / 52576, 525 ins, 362 del, 2577 sub ] exp/nnet5a_clean_100_gpu/decode_fglarge_test_clean/wer_13
%WER 6.76 [ 3556 / 52576, 517 ins, 400 del, 2639 sub ] exp/nnet5a_clean_100_gpu/decode_tglarge_test_clean/wer_13
%WER 8.56 [ 4503 / 52576, 524 ins, 624 del, 3355 sub ] exp/nnet5a_clean_100_gpu/decode_tgmed_test_clean/wer_13
%WER 9.66 [ 5081 / 52576, 522 ins, 752 del, 3807 sub ] exp/nnet5a_clean_100_gpu/decode_tgsmall_test_clean/wer_13
%WER 20.42 [ 10403 / 50948, 1167 ins, 1530 del, 7706 sub ] exp/nnet5a_clean_100_gpu/decode_fglarge_dev_other/wer_16
%WER 21.48 [ 10945 / 50948, 1195 ins, 1670 del, 8080 sub ] exp/nnet5a_clean_100_gpu/decode_tglarge_dev_other/wer_17
%WER 24.74 [ 12605 / 50948, 1008 ins, 2353 del, 9244 sub ] exp/nnet5a_clean_100_gpu/decode_tgmed_dev_other/wer_17
%WER 26.68 [ 13591 / 50948, 1094 ins, 2533 del, 9964 sub ] exp/nnet5a_clean_100_gpu/decode_tgsmall_dev_other/wer_15
%WER 22.47 [ 11762 / 52343, 1296 ins, 1690 del, 8776 sub ] exp/nnet5a_clean_100_gpu/decode_fglarge_test_other/wer_15
%WER 23.44 [ 12269 / 52343, 1343 ins, 1809 del, 9117 sub ] exp/nnet5a_clean_100_gpu/decode_tglarge_test_other/wer_15
%WER 26.59 [ 13919 / 52343, 1195 ins, 2493 del, 10231 sub ] exp/nnet5a_clean_100_gpu/decode_tgmed_test_other/wer_15
%WER 28.64 [ 14989 / 52343, 1170 ins, 2873 del, 10946 sub ] exp/nnet5a_clean_100_gpu/decode_tgsmall_test_other/wer_15
### p-norm DNN trained on "train-clean-100" + "train-clean-360"
%WER 5.27 [ 2865 / 54402, 425 ins, 273 del, 2167 sub ] exp/nnet6a_clean_460_gpu/decode_fglarge_dev_clean/wer_11
%WER 5.57 [ 3028 / 54402, 442 ins, 310 del, 2276 sub ] exp/nnet6a_clean_460_gpu/decode_tglarge_dev_clean/wer_11
%WER 7.16 [ 3895 / 54402, 411 ins, 537 del, 2947 sub ] exp/nnet6a_clean_460_gpu/decode_tgmed_dev_clean/wer_12
%WER 8.23 [ 4477 / 54402, 475 ins, 654 del, 3348 sub ] exp/nnet6a_clean_460_gpu/decode_tgsmall_dev_clean/wer_11
%WER 5.78 [ 3038 / 52576, 483 ins, 293 del, 2262 sub ] exp/nnet6a_clean_460_gpu/decode_fglarge_test_clean/wer_11
%WER 6.18 [ 3248 / 52576, 505 ins, 330 del, 2413 sub ] exp/nnet6a_clean_460_gpu/decode_tglarge_test_clean/wer_11
%WER 7.74 [ 4067 / 52576, 450 ins, 599 del, 3018 sub ] exp/nnet6a_clean_460_gpu/decode_tgmed_test_clean/wer_13
%WER 8.71 [ 4581 / 52576, 510 ins, 628 del, 3443 sub ] exp/nnet6a_clean_460_gpu/decode_tgsmall_test_clean/wer_11
%WER 17.67 [ 9000 / 50948, 979 ins, 1217 del, 6804 sub ] exp/nnet6a_clean_460_gpu/decode_fglarge_dev_other/wer_14
%WER 18.58 [ 9468 / 50948, 999 ins, 1410 del, 7059 sub ] exp/nnet6a_clean_460_gpu/decode_tglarge_dev_other/wer_15
%WER 21.89 [ 11155 / 50948, 1016 ins, 1739 del, 8400 sub ] exp/nnet6a_clean_460_gpu/decode_tgmed_dev_other/wer_13
%WER 23.75 [ 12098 / 50948, 983 ins, 2084 del, 9031 sub ] exp/nnet6a_clean_460_gpu/decode_tgsmall_dev_other/wer_13
%WER 19.12 [ 10008 / 52343, 1062 ins, 1448 del, 7498 sub ] exp/nnet6a_clean_460_gpu/decode_fglarge_test_other/wer_14
%WER 20.07 [ 10507 / 52343, 1114 ins, 1548 del, 7845 sub ] exp/nnet6a_clean_460_gpu/decode_tglarge_test_other/wer_14
%WER 23.22 [ 12155 / 52343, 1037 ins, 2151 del, 8967 sub ] exp/nnet6a_clean_460_gpu/decode_tgmed_test_other/wer_14
%WER 25.34 [ 13265 / 52343, 990 ins, 2567 del, 9708 sub ] exp/nnet6a_clean_460_gpu/decode_tgsmall_test_other/wer_14
### p-norm DNN trained on "train-clean-100" + "train-clean-360" + "train-other-500"
%WER 4.90 [ 2665 / 54402, 400 ins, 258 del, 2007 sub ] exp/nnet7a_960_gpu/decode_fglarge_dev_clean/wer_12
%WER 5.14 [ 2795 / 54402, 404 ins, 286 del, 2105 sub ] exp/nnet7a_960_gpu/decode_tglarge_dev_clean/wer_12
%WER 6.57 [ 3572 / 54402, 402 ins, 475 del, 2695 sub ] exp/nnet7a_960_gpu/decode_tgmed_dev_clean/wer_12
%WER 7.54 [ 4103 / 54402, 425 ins, 598 del, 3080 sub ] exp/nnet7a_960_gpu/decode_tgsmall_dev_clean/wer_12
%WER 5.49 [ 2886 / 52576, 452 ins, 292 del, 2142 sub ] exp/nnet7a_960_gpu/decode_fglarge_test_clean/wer_13
%WER 5.74 [ 3017 / 52576, 468 ins, 317 del, 2232 sub ] exp/nnet7a_960_gpu/decode_tglarge_test_clean/wer_12
%WER 7.21 [ 3789 / 52576, 481 ins, 478 del, 2830 sub ] exp/nnet7a_960_gpu/decode_tgmed_test_clean/wer_12
%WER 8.01 [ 4213 / 52576, 503 ins, 543 del, 3167 sub ] exp/nnet7a_960_gpu/decode_tgsmall_test_clean/wer_11
%WER 12.98 [ 6614 / 50948, 788 ins, 825 del, 5001 sub ] exp/nnet7a_960_gpu/decode_fglarge_dev_other/wer_13
%WER 13.89 [ 7078 / 50948, 883 ins, 844 del, 5351 sub ] exp/nnet7a_960_gpu/decode_tglarge_dev_other/wer_12
%WER 16.72 [ 8520 / 50948, 808 ins, 1299 del, 6413 sub ] exp/nnet7a_960_gpu/decode_tgmed_dev_other/wer_13
%WER 18.51 [ 9433 / 50948, 806 ins, 1609 del, 7018 sub ] exp/nnet7a_960_gpu/decode_tgsmall_dev_other/wer_13
%WER 13.97 [ 7311 / 52343, 858 ins, 958 del, 5495 sub ] exp/nnet7a_960_gpu/decode_fglarge_test_other/wer_13
%WER 14.77 [ 7733 / 52343, 914 ins, 989 del, 5830 sub ] exp/nnet7a_960_gpu/decode_tglarge_test_other/wer_12
%WER 17.58 [ 9204 / 52343, 867 ins, 1415 del, 6922 sub ] exp/nnet7a_960_gpu/decode_tgmed_test_other/wer_12
%WER 19.41 [ 10158 / 52343, 888 ins, 1689 del, 7581 sub ] exp/nnet7a_960_gpu/decode_tgsmall_test_other/wer_12
### online-nnet2 results with a model trained on all(960h) of the training data
### Note: these results are now superseded by the multi-splice (_ms_) results below.
### Be careful when comparing, as the _ms_ results don't yet have the _fglarge tests.
%WER 4.90 [ 2663 / 54402, 388 ins, 273 del, 2002 sub ] exp/nnet2_online/nnet_a_online/decode_dev_clean_fglarge/wer_13
%WER 5.19 [ 2822 / 54402, 406 ins, 311 del, 2105 sub ] exp/nnet2_online/nnet_a_online/decode_dev_clean_tglarge/wer_13
%WER 6.60 [ 3593 / 54402, 457 ins, 426 del, 2710 sub ] exp/nnet2_online/nnet_a_online/decode_dev_clean_tgmed/wer_11
%WER 7.46 [ 4059 / 54402, 434 ins, 574 del, 3051 sub ] exp/nnet2_online/nnet_a_online/decode_dev_clean_tgsmall/wer_12
%WER 5.52 [ 2900 / 52576, 456 ins, 279 del, 2165 sub ] exp/nnet2_online/nnet_a_online/decode_test_clean_fglarge/wer_12
%WER 5.71 [ 3002 / 52576, 452 ins, 322 del, 2228 sub ] exp/nnet2_online/nnet_a_online/decode_test_clean_tglarge/wer_12
%WER 7.17 [ 3770 / 52576, 486 ins, 444 del, 2840 sub ] exp/nnet2_online/nnet_a_online/decode_test_clean_tgmed/wer_11
%WER 7.97 [ 4188 / 52576, 459 ins, 562 del, 3167 sub ] exp/nnet2_online/nnet_a_online/decode_test_clean_tgsmall/wer_12
%WER 13.59 [ 6926 / 50948, 821 ins, 892 del, 5213 sub ] exp/nnet2_online/nnet_a_online/decode_dev_other_fglarge/wer_14
%WER 14.06 [ 7165 / 50948, 865 ins, 911 del, 5389 sub ] exp/nnet2_online/nnet_a_online/decode_dev_other_tglarge/wer_13
%WER 16.77 [ 8546 / 50948, 828 ins, 1299 del, 6419 sub ] exp/nnet2_online/nnet_a_online/decode_dev_other_tgmed/wer_13
%WER 18.46 [ 9405 / 50948, 797 ins, 1580 del, 7028 sub ] exp/nnet2_online/nnet_a_online/decode_dev_other_tgsmall/wer_13
%WER 13.79 [ 7217 / 52343, 866 ins, 894 del, 5457 sub ] exp/nnet2_online/nnet_a_online/decode_test_other_fglarge/wer_12
%WER 14.39 [ 7532 / 52343, 895 ins, 959 del, 5678 sub ] exp/nnet2_online/nnet_a_online/decode_test_other_tglarge/wer_12
%WER 17.16 [ 8982 / 52343, 855 ins, 1421 del, 6706 sub ] exp/nnet2_online/nnet_a_online/decode_test_other_tgmed/wer_12
%WER 18.90 [ 9891 / 52343, 798 ins, 1786 del, 7307 sub ] exp/nnet2_online/nnet_a_online/decode_test_other_tgsmall/wer_13
# RNNLM rescoring of tri6b (faster-rnnlm hidden=150 direct=4.0Gb, Hierarchical Softmax)
%WER 7.39 [ 4023 / 54402, 540 ins, 444 del, 3039 sub ] exp/tri6b/decode_tglarge_dev_clean/wer_13_1.0
%WER 7.03 [ 3823 / 54402, 608 ins, 343 del, 2872 sub ] exp/tri6b/decode_tglarge_dev_clean_faster-rnnlm_h150-me5-1000_L0.25/wer_13_0.5
%WER 7.03 [ 3827 / 54402, 606 ins, 320 del, 2901 sub ] exp/tri6b/decode_tglarge_dev_clean_faster-rnnlm_h150-me5-1000_L0.5/wer_14_0.5
%WER 7.25 [ 3946 / 54402, 564 ins, 368 del, 3014 sub ] exp/tri6b/decode_tglarge_dev_clean_faster-rnnlm_h150-me5-1000_L0.75/wer_14_1.0
%WER 21.31 [ 10858 / 50948, 1525 ins, 1151 del, 8182 sub ] exp/tri6b/decode_tglarge_dev_other/wer_17_0.0
%WER 20.62 [ 10504 / 50948, 1377 ins, 1180 del, 7947 sub ] exp/tri6b/decode_tglarge_dev_other_faster-rnnlm_h150-me5-1000_L0.25/wer_15_0.5
%WER 20.64 [ 10515 / 50948, 1253 ins, 1313 del, 7949 sub ] exp/tri6b/decode_tglarge_dev_other_faster-rnnlm_h150-me5-1000_L0.5/wer_16_1.0
%WER 20.91 [ 10652 / 50948, 1344 ins, 1233 del, 8075 sub ] exp/tri6b/decode_tglarge_dev_other_faster-rnnlm_h150-me5-1000_L0.75/wer_15_1.0
%WER 9.21 [ 5012 / 54402, 703 ins, 510 del, 3799 sub ] exp/tri6b/decode_tgmed_dev_clean/wer_14_0.0
%WER 7.99 [ 4345 / 54402, 554 ins, 487 del, 3304 sub ] exp/tri6b/decode_tgmed_dev_clean_faster-rnnlm_h150-me5-1000_L0.25/wer_15_0.5
%WER 7.68 [ 4177 / 54402, 596 ins, 414 del, 3167 sub ] exp/tri6b/decode_tgmed_dev_clean_faster-rnnlm_h150-me5-1000_L0.5/wer_14_0.5
%WER 7.70 [ 4190 / 54402, 582 ins, 422 del, 3186 sub ] exp/tri6b/decode_tgmed_dev_clean_faster-rnnlm_h150-me5-1000_L0.75/wer_13_1.0
%WER 24.27 [ 12365 / 50948, 1365 ins, 1591 del, 9409 sub ] exp/tri6b/decode_tgmed_dev_other/wer_17_0.0
%WER 22.51 [ 11468 / 50948, 1496 ins, 1235 del, 8737 sub ] exp/tri6b/decode_tgmed_dev_other_faster-rnnlm_h150-me5-1000_L0.25/wer_15_0.0
%WER 22.11 [ 11267 / 50948, 1494 ins, 1163 del, 8610 sub ] exp/tri6b/decode_tgmed_dev_other_faster-rnnlm_h150-me5-1000_L0.5/wer_16_0.0
%WER 22.10 [ 11262 / 50948, 1532 ins, 1131 del, 8599 sub ] exp/tri6b/decode_tgmed_dev_other_faster-rnnlm_h150-me5-1000_L0.75/wer_16_0.0
%WER 10.50 [ 5711 / 54402, 693 ins, 674 del, 4344 sub ] exp/tri6b/decode_tgsmall_dev_clean/wer_15_0.0
%WER 8.53 [ 4641 / 54402, 582 ins, 555 del, 3504 sub ] exp/tri6b/decode_tgsmall_dev_clean_faster-rnnlm_h150-me5-1000_L0.25/wer_14_0.5
%WER 8.09 [ 4400 / 54402, 605 ins, 469 del, 3326 sub ] exp/tri6b/decode_tgsmall_dev_clean_faster-rnnlm_h150-me5-1000_L0.5/wer_14_0.5
%WER 8.02 [ 4363 / 54402, 594 ins, 460 del, 3309 sub ] exp/tri6b/decode_tgsmall_dev_clean_faster-rnnlm_h150-me5-1000_L0.75/wer_13_1.0
%WER 26.22 [ 13358 / 50948, 1330 ins, 1955 del, 10073 sub ] exp/tri6b/decode_tgsmall_dev_other/wer_17_0.0
%WER 23.95 [ 12202 / 50948, 1523 ins, 1381 del, 9298 sub ] exp/tri6b/decode_tgsmall_dev_other_faster-rnnlm_h150-me5-1000_L0.25/wer_14_0.0
%WER 23.22 [ 11828 / 50948, 1553 ins, 1247 del, 9028 sub ] exp/tri6b/decode_tgsmall_dev_other_faster-rnnlm_h150-me5-1000_L0.5/wer_14_0.0
%WER 23.22 [ 11832 / 50948, 1435 ins, 1376 del, 9021 sub ] exp/tri6b/decode_tgsmall_dev_other_faster-rnnlm_h150-me5-1000_L0.75/wer_15_0.5
# RNNLM rescoring of tri6b (faster-rnnlm hidden=150 direct=1.6Gb Noise contrastive Estimation)
%WER 7.39 [ 4023 / 54402, 540 ins, 444 del, 3039 sub ] exp/tri6b/decode_tglarge_dev_clean/wer_13_1.0
%WER 7.05 [ 3835 / 54402, 487 ins, 447 del, 2901 sub ] exp/tri6b/decode_tglarge_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.25/wer_15_1.0
%WER 6.84 [ 3723 / 54402, 524 ins, 394 del, 2805 sub ] exp/tri6b/decode_tglarge_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.5/wer_13_1.0
%WER 6.92 [ 3766 / 54402, 564 ins, 376 del, 2826 sub ] exp/tri6b/decode_tglarge_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.75/wer_12_1.0
%WER 21.31 [ 10858 / 50948, 1525 ins, 1151 del, 8182 sub ] exp/tri6b/decode_tglarge_dev_other/wer_17_0.0
%WER 20.90 [ 10648 / 50948, 1404 ins, 1227 del, 8017 sub ] exp/tri6b/decode_tglarge_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.25/wer_15_0.5
%WER 20.70 [ 10544 / 50948, 1271 ins, 1364 del, 7909 sub ] exp/tri6b/decode_tglarge_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.5/wer_15_1.0
%WER 20.82 [ 10605 / 50948, 1295 ins, 1347 del, 7963 sub ] exp/tri6b/decode_tglarge_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.75/wer_15_1.0
%WER 9.21 [ 5012 / 54402, 703 ins, 510 del, 3799 sub ] exp/tri6b/decode_tgmed_dev_clean/wer_14_0.0
%WER 8.01 [ 4360 / 54402, 669 ins, 402 del, 3289 sub ] exp/tri6b/decode_tgmed_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.25/wer_14_0.0
%WER 7.46 [ 4056 / 54402, 584 ins, 422 del, 3050 sub ] exp/tri6b/decode_tgmed_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.5/wer_14_0.5
%WER 7.28 [ 3962 / 54402, 536 ins, 451 del, 2975 sub ] exp/tri6b/decode_tgmed_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.75/wer_14_1.0
%WER 24.27 [ 12365 / 50948, 1365 ins, 1591 del, 9409 sub ] exp/tri6b/decode_tgmed_dev_other/wer_17_0.0
%WER 22.82 [ 11628 / 50948, 1530 ins, 1244 del, 8854 sub ] exp/tri6b/decode_tgmed_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.25/wer_15_0.0
%WER 22.21 [ 11315 / 50948, 1554 ins, 1152 del, 8609 sub ] exp/tri6b/decode_tgmed_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.5/wer_15_0.0
%WER 22.01 [ 11213 / 50948, 1609 ins, 1086 del, 8518 sub ] exp/tri6b/decode_tgmed_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.75/wer_15_0.0
%WER 10.50 [ 5711 / 54402, 693 ins, 674 del, 4344 sub ] exp/tri6b/decode_tgsmall_dev_clean/wer_15_0.0
%WER 8.56 [ 4659 / 54402, 677 ins, 467 del, 3515 sub ] exp/tri6b/decode_tgsmall_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.25/wer_14_0.0
%WER 7.81 [ 4250 / 54402, 657 ins, 387 del, 3206 sub ] exp/tri6b/decode_tgsmall_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.5/wer_14_0.0
%WER 7.58 [ 4125 / 54402, 618 ins, 406 del, 3101 sub ] exp/tri6b/decode_tgsmall_dev_clean_faster-rnnlm_h150-me3-400-nce20_L0.75/wer_13_0.5
%WER 26.22 [ 13358 / 50948, 1330 ins, 1955 del, 10073 sub ] exp/tri6b/decode_tgsmall_dev_other/wer_17_0.0
%WER 24.07 [ 12264 / 50948, 1482 ins, 1435 del, 9347 sub ] exp/tri6b/decode_tgsmall_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.25/wer_15_0.0
%WER 23.15 [ 11797 / 50948, 1526 ins, 1276 del, 8995 sub ] exp/tri6b/decode_tgsmall_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.5/wer_15_0.0
%WER 22.92 [ 11677 / 50948, 1544 ins, 1241 del, 8892 sub ] exp/tri6b/decode_tgsmall_dev_other_faster-rnnlm_h150-me3-400-nce20_L0.75/wer_16_0.0
## Multi-splice version of online recipe.
# for x in exp/nnet2_online/nnet_ms_a/decode_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 4.72 [ 2568 / 54402, 390 ins, 258 del, 1920 sub ] exp/nnet2_online/nnet_ms_i2/decode_dev_clean_tglarge/wer_12
%WER 5.90 [ 3212 / 54402, 345 ins, 441 del, 2426 sub ] exp/nnet2_online/nnet_ms_i2/decode_dev_clean_tgmed/wer_14
%WER 6.64 [ 3612 / 54402, 401 ins, 479 del, 2732 sub ] exp/nnet2_online/nnet_ms_i2/decode_dev_clean_tgsmall/wer_12
%WER 13.11 [ 6680 / 50948, 797 ins, 866 del, 5017 sub ] exp/nnet2_online/nnet_ms_i2/decode_dev_other_tglarge/wer_15
%WER 15.56 [ 7925 / 50948, 727 ins, 1261 del, 5937 sub ] exp/nnet2_online/nnet_ms_i2/decode_dev_other_tgmed/wer_15
%WER 17.10 [ 8714 / 50948, 733 ins, 1510 del, 6471 sub ] exp/nnet2_online/nnet_ms_i2/decode_dev_other_tgsmall/wer_15
# for x in exp/nnet2_online/nnet_ms_a_online/decode_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 4.83 [ 2629 / 54402, 393 ins, 264 del, 1972 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tglarge/wer_13
%WER 5.01 [ 2726 / 54402, 402 ins, 270 del, 2054 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tglarge_utt/wer_13
%WER 4.87 [ 2647 / 54402, 386 ins, 290 del, 1971 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tglarge_utt_offline/wer_14
%WER 6.05 [ 3294 / 54402, 409 ins, 392 del, 2493 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgmed/wer_12
%WER 6.30 [ 3428 / 54402, 389 ins, 434 del, 2605 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgmed_utt/wer_13
%WER 6.09 [ 3311 / 54402, 393 ins, 417 del, 2501 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgmed_utt_offline/wer_13
%WER 6.87 [ 3740 / 54402, 390 ins, 547 del, 2803 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgsmall/wer_13
%WER 7.21 [ 3921 / 54402, 440 ins, 535 del, 2946 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgsmall_utt/wer_12
%WER 6.95 [ 3783 / 54402, 415 ins, 543 del, 2825 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgsmall_utt_offline/wer_13
%WER 13.21 [ 6732 / 50948, 812 ins, 852 del, 5068 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tglarge/wer_14
%WER 14.24 [ 7254 / 50948, 884 ins, 959 del, 5411 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tglarge_utt/wer_15
%WER 13.63 [ 6945 / 50948, 890 ins, 856 del, 5199 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tglarge_utt_offline/wer_14
%WER 15.69 [ 7996 / 50948, 800 ins, 1189 del, 6007 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgmed/wer_14
%WER 16.63 [ 8473 / 50948, 809 ins, 1317 del, 6347 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgmed_utt/wer_15
%WER 16.09 [ 8197 / 50948, 872 ins, 1130 del, 6195 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgmed_utt_offline/wer_13
%WER 17.15 [ 8736 / 50948, 756 ins, 1424 del, 6556 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgsmall/wer_14
%WER 18.23 [ 9288 / 50948, 782 ins, 1585 del, 6921 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgsmall_utt/wer_15
%WER 17.54 [ 8936 / 50948, 813 ins, 1425 del, 6698 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgsmall_utt_offline/wer_14
## Multi-splice version of online recipe (5/16/2016).
# for x in exp/nnet2_online/nnet_ms_a/decode_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 4.46 [ 2429 / 54402, 311 ins, 284 del, 1834 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_clean_fglarge/wer_13_1.0
%WER 4.64 [ 2522 / 54402, 362 ins, 251 del, 1909 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_clean_tglarge/wer_12_0.5
%WER 5.86 [ 3187 / 54402, 400 ins, 357 del, 2430 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_clean_tgmed/wer_13_0.0
%WER 6.60 [ 3592 / 54402, 450 ins, 403 del, 2739 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_clean_tgsmall/wer_12_0.0
%WER 12.31 [ 6274 / 50948, 742 ins, 784 del, 4748 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_other_fglarge/wer_16_0.5
%WER 12.87 [ 6557 / 50948, 774 ins, 850 del, 4933 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_other_tglarge/wer_15_0.5
%WER 15.25 [ 7770 / 50948, 871 ins, 1074 del, 5825 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_other_tgmed/wer_16_0.0
%WER 16.55 [ 8434 / 50948, 832 ins, 1280 del, 6322 sub ] exp/nnet2_online/nnet_ms_a/decode_dev_other_tgsmall/wer_16_0.0
%WER 4.99 [ 2624 / 52576, 388 ins, 256 del, 1980 sub ] exp/nnet2_online/nnet_ms_a/decode_test_clean_fglarge/wer_13_0.5
%WER 5.15 [ 2709 / 52576, 386 ins, 284 del, 2039 sub ] exp/nnet2_online/nnet_ms_a/decode_test_clean_tglarge/wer_13_0.5
%WER 6.25 [ 3285 / 52576, 422 ins, 357 del, 2506 sub ] exp/nnet2_online/nnet_ms_a/decode_test_clean_tgmed/wer_13_0.0
%WER 7.07 [ 3717 / 52576, 455 ins, 456 del, 2806 sub ] exp/nnet2_online/nnet_ms_a/decode_test_clean_tgsmall/wer_13_0.0
%WER 12.89 [ 6748 / 52343, 878 ins, 769 del, 5101 sub ] exp/nnet2_online/nnet_ms_a/decode_test_other_fglarge/wer_16_0.0
%WER 13.32 [ 6972 / 52343, 940 ins, 770 del, 5262 sub ] exp/nnet2_online/nnet_ms_a/decode_test_other_tglarge/wer_14_0.0
%WER 15.82 [ 8281 / 52343, 886 ins, 1197 del, 6198 sub ] exp/nnet2_online/nnet_ms_a/decode_test_other_tgmed/wer_15_0.0
%WER 17.09 [ 8948 / 52343, 863 ins, 1383 del, 6702 sub ] exp/nnet2_online/nnet_ms_a/decode_test_other_tgsmall/wer_15_0.0
# for x in exp/nnet2_online/nnet_ms_a_online/decode_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 4.53 [ 2466 / 54402, 318 ins, 295 del, 1853 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_fglarge/wer_14_1.0
%WER 4.76 [ 2592 / 54402, 338 ins, 286 del, 1968 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_fglarge_utt/wer_13_1.0
%WER 4.57 [ 2488 / 54402, 330 ins, 285 del, 1873 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_fglarge_utt_offline/wer_13_1.0
%WER 4.71 [ 2562 / 54402, 392 ins, 236 del, 1934 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tglarge/wer_14_0.0
%WER 4.90 [ 2665 / 54402, 352 ins, 280 del, 2033 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tglarge_utt/wer_14_0.5
%WER 4.72 [ 2570 / 54402, 357 ins, 273 del, 1940 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tglarge_utt_offline/wer_14_0.5
%WER 5.87 [ 3196 / 54402, 419 ins, 340 del, 2437 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgmed/wer_12_0.0
%WER 6.11 [ 3326 / 54402, 385 ins, 396 del, 2545 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgmed_utt/wer_12_0.5
%WER 5.99 [ 3258 / 54402, 382 ins, 392 del, 2484 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgmed_utt_offline/wer_12_0.5
%WER 6.58 [ 3581 / 54402, 472 ins, 379 del, 2730 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgsmall/wer_11_0.0
%WER 6.89 [ 3746 / 54402, 475 ins, 405 del, 2866 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgsmall_utt/wer_12_0.0
%WER 6.69 [ 3637 / 54402, 480 ins, 383 del, 2774 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_clean_tgsmall_utt_offline/wer_11_0.0
%WER 12.67 [ 6456 / 50948, 774 ins, 771 del, 4911 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_fglarge/wer_16_0.5
%WER 13.73 [ 6993 / 50948, 785 ins, 922 del, 5286 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_fglarge_utt/wer_14_1.0
%WER 12.97 [ 6609 / 50948, 797 ins, 801 del, 5011 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_fglarge_utt_offline/wer_16_0.5
%WER 13.09 [ 6670 / 50948, 800 ins, 826 del, 5044 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tglarge/wer_15_0.5
%WER 14.27 [ 7270 / 50948, 909 ins, 869 del, 5492 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tglarge_utt/wer_14_0.5
%WER 13.46 [ 6859 / 50948, 828 ins, 845 del, 5186 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tglarge_utt_offline/wer_15_0.5
%WER 15.27 [ 7782 / 50948, 874 ins, 1051 del, 5857 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgmed/wer_16_0.0
%WER 16.41 [ 8359 / 50948, 949 ins, 1135 del, 6275 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgmed_utt/wer_16_0.0
%WER 15.56 [ 7926 / 50948, 893 ins, 1051 del, 5982 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgmed_utt_offline/wer_16_0.0
%WER 16.49 [ 8402 / 50948, 855 ins, 1210 del, 6337 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgsmall/wer_15_0.0
%WER 17.80 [ 9068 / 50948, 969 ins, 1260 del, 6839 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgsmall_utt/wer_15_0.0
%WER 16.97 [ 8647 / 50948, 845 ins, 1324 del, 6478 sub ] exp/nnet2_online/nnet_ms_a_online/decode_dev_other_tgsmall_utt_offline/wer_17_0.0
%WER 5.05 [ 2654 / 52576, 411 ins, 239 del, 2004 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_fglarge/wer_12_0.5
%WER 5.24 [ 2755 / 52576, 365 ins, 312 del, 2078 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_fglarge_utt/wer_13_1.0
%WER 5.09 [ 2676 / 52576, 405 ins, 241 del, 2030 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_fglarge_utt_offline/wer_13_0.5
%WER 5.22 [ 2744 / 52576, 393 ins, 282 del, 2069 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tglarge/wer_13_0.5
%WER 5.38 [ 2826 / 52576, 413 ins, 284 del, 2129 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tglarge_utt/wer_13_0.5
%WER 5.24 [ 2757 / 52576, 453 ins, 229 del, 2075 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tglarge_utt_offline/wer_13_0.0
%WER 6.26 [ 3289 / 52576, 436 ins, 345 del, 2508 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tgmed/wer_13_0.0
%WER 6.54 [ 3441 / 52576, 435 ins, 381 del, 2625 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tgmed_utt/wer_14_0.0
%WER 6.28 [ 3303 / 52576, 426 ins, 359 del, 2518 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tgmed_utt_offline/wer_14_0.0
%WER 7.06 [ 3711 / 52576, 446 ins, 474 del, 2791 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tgsmall/wer_14_0.0
%WER 7.31 [ 3845 / 52576, 510 ins, 426 del, 2909 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tgsmall_utt/wer_12_0.0
%WER 7.08 [ 3723 / 52576, 460 ins, 445 del, 2818 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_clean_tgsmall_utt_offline/wer_13_0.0
%WER 13.17 [ 6891 / 52343, 936 ins, 713 del, 5242 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_fglarge/wer_14_0.0
%WER 14.20 [ 7432 / 52343, 832 ins, 983 del, 5617 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_fglarge_utt/wer_15_0.5
%WER 13.26 [ 6939 / 52343, 837 ins, 860 del, 5242 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_fglarge_utt_offline/wer_14_0.5
%WER 13.53 [ 7080 / 52343, 952 ins, 779 del, 5349 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tglarge/wer_14_0.0
%WER 14.77 [ 7730 / 52343, 877 ins, 1056 del, 5797 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tglarge_utt/wer_15_0.5
%WER 13.74 [ 7192 / 52343, 871 ins, 920 del, 5401 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tglarge_utt_offline/wer_14_0.5
%WER 15.78 [ 8259 / 52343, 898 ins, 1170 del, 6191 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tgmed/wer_15_0.0
%WER 16.97 [ 8884 / 52343, 939 ins, 1304 del, 6641 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tgmed_utt/wer_16_0.0
%WER 16.01 [ 8380 / 52343, 877 ins, 1210 del, 6293 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tgmed_utt_offline/wer_16_0.0
%WER 16.98 [ 8889 / 52343, 900 ins, 1283 del, 6706 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tgsmall/wer_14_0.0
%WER 18.21 [ 9533 / 52343, 966 ins, 1398 del, 7169 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tgsmall_utt/wer_14_0.0
%WER 17.29 [ 9050 / 52343, 894 ins, 1391 del, 6765 sub ] exp/nnet2_online/nnet_ms_a_online/decode_test_other_tgsmall_utt_offline/wer_15_0.0
## Note: this learning rate is the effective learning rate; it gets multiplied by the num-jobs.
# for x in exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch*{clean,other}*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 5.92 [ 3221 / 54402, 352 ins, 439 del, 2430 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch0_dev_clean_tgmed/wer_14
%WER 6.63 [ 3605 / 54402, 399 ins, 481 del, 2725 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch0_dev_clean_tgsmall/wer_12
%WER 4.44 [ 2416 / 54402, 385 ins, 204 del, 1827 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch1_dev_clean_tglarge/wer_14
%WER 5.52 [ 3001 / 54402, 360 ins, 340 del, 2301 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch1_dev_clean_tgmed/wer_15
%WER 6.22 [ 3384 / 54402, 388 ins, 411 del, 2585 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch1_dev_clean_tgsmall/wer_14
%WER 4.39 [ 2386 / 54402, 368 ins, 208 del, 1810 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch2_dev_clean_tglarge/wer_15 **
%WER 5.41 [ 2945 / 54402, 338 ins, 339 del, 2268 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch2_dev_clean_tgmed/wer_16
%WER 6.13 [ 3333 / 54402, 371 ins, 410 del, 2552 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch2_dev_clean_tgsmall/wer_15
%WER 4.39 [ 2387 / 54402, 377 ins, 199 del, 1811 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch3_dev_clean_tglarge/wer_14
%WER 5.36 [ 2918 / 54402, 328 ins, 338 del, 2252 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch3_dev_clean_tgmed/wer_17
%WER 6.08 [ 3305 / 54402, 369 ins, 396 del, 2540 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch3_dev_clean_tgsmall/wer_15
%WER 4.40 [ 2395 / 54402, 375 ins, 200 del, 1820 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch4_dev_clean_tglarge/wer_14
%WER 5.35 [ 2909 / 54402, 328 ins, 339 del, 2242 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch4_dev_clean_tgmed/wer_17
%WER 6.05 [ 3291 / 54402, 384 ins, 381 del, 2526 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch4_dev_clean_tgsmall/wer_14
%WER 13.45 [ 6850 / 50948, 808 ins, 876 del, 5166 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch0_dev_other_tglarge/wer_15
%WER 15.65 [ 7975 / 50948, 714 ins, 1311 del, 5950 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch0_dev_other_tgmed/wer_16
%WER 17.12 [ 8722 / 50948, 739 ins, 1489 del, 6494 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch0_dev_other_tgsmall/wer_15
%WER 12.84 [ 6544 / 50948, 877 ins, 703 del, 4964 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch1_dev_other_tglarge/wer_16
%WER 14.87 [ 7578 / 50948, 742 ins, 1102 del, 5734 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch1_dev_other_tgmed/wer_18
%WER 16.25 [ 8277 / 50948, 823 ins, 1171 del, 6283 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch1_dev_other_tgsmall/wer_15
%WER 12.80 [ 6522 / 50948, 869 ins, 698 del, 4955 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch2_dev_other_tglarge/wer_17 **
%WER 14.80 [ 7542 / 50948, 774 ins, 1034 del, 5734 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch2_dev_other_tgmed/wer_17
%WER 16.14 [ 8225 / 50948, 763 ins, 1242 del, 6220 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch2_dev_other_tgsmall/wer_17
%WER 12.82 [ 6531 / 50948, 871 ins, 710 del, 4950 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch3_dev_other_tglarge/wer_18
%WER 14.82 [ 7549 / 50948, 818 ins, 958 del, 5773 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch3_dev_other_tgmed/wer_16
%WER 16.10 [ 8204 / 50948, 795 ins, 1165 del, 6244 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch3_dev_other_tgsmall/wer_16
%WER 12.85 [ 6549 / 50948, 902 ins, 672 del, 4975 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch4_dev_other_tglarge/wer_17
%WER 14.80 [ 7540 / 50948, 800 ins, 1025 del, 5715 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch4_dev_other_tgmed/wer_18
%WER 16.10 [ 8201 / 50948, 789 ins, 1240 del, 6172 sub ] exp/nnet2_online/nnet_ms_a_smbr_0.000005/decode_epoch4_dev_other_tgsmall/wer_18
## Results with a SAT model, trained on the "train-clean-100" and pronunciation probabilities estimated on the training data
# for x in exp/tri4b/decode_pp_*; do grep WER $x/wer_* | utils/best_wer.sh | egrep -v '\.si' ; done
%WER 8.05 [ 4382 / 54402, 583 ins, 520 del, 3279 sub ] exp/tri4b/decode_pp_fglarge_dev_clean/wer_17
%WER 29.07 [ 14813 / 50948, 1411 ins, 2373 del, 11029 sub ] exp/tri4b/decode_pp_fglarge_dev_other/wer_18
%WER 9.19 [ 4833 / 52576, 621 ins, 579 del, 3633 sub ] exp/tri4b/decode_pp_fglarge_test_clean/wer_20
%WER 31.28 [ 16371 / 52343, 1505 ins, 2773 del, 12093 sub ] exp/tri4b/decode_pp_fglarge_test_other/wer_17
%WER 8.53 [ 4639 / 54402, 635 ins, 516 del, 3488 sub ] exp/tri4b/decode_pp_tglarge_dev_clean/wer_15
%WER 29.98 [ 15274 / 50948, 1501 ins, 2380 del, 11393 sub ] exp/tri4b/decode_pp_tglarge_dev_other/wer_18
%WER 9.45 [ 4969 / 52576, 661 ins, 577 del, 3731 sub ] exp/tri4b/decode_pp_tglarge_test_clean/wer_18
%WER 32.14 [ 16824 / 52343, 1649 ins, 2804 del, 12371 sub ] exp/tri4b/decode_pp_tglarge_test_other/wer_17
%WER 10.47 [ 5694 / 54402, 615 ins, 793 del, 4286 sub ] exp/tri4b/decode_pp_tgmed_dev_clean/wer_15
%WER 32.97 [ 16795 / 50943, 1416 ins, 2874 del, 12505 sub ] [PARTIAL] exp/tri4b/decode_pp_tgmed_dev_other/wer_16
%WER 11.67 [ 6133 / 52576, 685 ins, 831 del, 4617 sub ] exp/tri4b/decode_pp_tgmed_test_clean/wer_16
%WER 34.97 [ 18303 / 52343, 1409 ins, 3660 del, 13234 sub ] exp/tri4b/decode_pp_tgmed_test_other/wer_17
%WER 11.93 [ 6490 / 54402, 641 ins, 1017 del, 4832 sub ] exp/tri4b/decode_pp_tgsmall_dev_clean/wer_15
%WER 34.98 [ 17821 / 50948, 1396 ins, 3344 del, 13081 sub ] exp/tri4b/decode_pp_tgsmall_dev_other/wer_16
%WER 13.07 [ 6874 / 52576, 698 ins, 986 del, 5190 sub ] exp/tri4b/decode_pp_tgsmall_test_clean/wer_15
%WER 36.83 [ 19276 / 52343, 1261 ins, 4092 del, 13923 sub ] exp/tri4b/decode_pp_tgsmall_test_other/wer_17
## Multi-splice version of the online recipe, using pronunciation probabilities estimated on training data
# for x in exp/nnet2_online/nnet_ms_a/decode_pp_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 4.43 [ 2411 / 54402, 339 ins, 258 del, 1814 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_clean_fglarge/wer_14
%WER 4.70 [ 2555 / 54402, 388 ins, 246 del, 1921 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_clean_tglarge/wer_11
%WER 5.86 [ 3186 / 54402, 338 ins, 449 del, 2399 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_clean_tgmed/wer_14
%WER 6.59 [ 3587 / 54402, 381 ins, 486 del, 2720 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_clean_tgsmall/wer_12
%WER 12.50 [ 6371 / 50948, 702 ins, 898 del, 4771 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_other_fglarge/wer_16
%WER 13.05 [ 6648 / 50948, 755 ins, 916 del, 4977 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_other_tglarge/wer_15
%WER 15.57 [ 7935 / 50948, 688 ins, 1327 del, 5920 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_other_tgmed/wer_15
%WER 17.08 [ 8702 / 50948, 694 ins, 1567 del, 6441 sub ] exp/nnet2_online/nnet_ms_a/decode_pp_dev_other_tgsmall/wer_15
# for x in exp/nnet2_online/nnet_ms_a_online/decode_pp_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 4.50 [ 2448 / 54402, 346 ins, 262 del, 1840 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_fglarge/wer_14
%WER 4.66 [ 2537 / 54402, 374 ins, 243 del, 1920 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_fglarge_utt/wer_12
%WER 4.52 [ 2461 / 54402, 359 ins, 252 del, 1850 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_fglarge_utt_offline/wer_13
%WER 4.74 [ 2581 / 54402, 375 ins, 272 del, 1934 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tglarge/wer_12
%WER 4.83 [ 2625 / 54402, 358 ins, 279 del, 1988 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tglarge_utt/wer_13
%WER 4.72 [ 2567 / 54402, 361 ins, 283 del, 1923 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tglarge_utt_offline/wer_13
%WER 5.85 [ 3184 / 54402, 343 ins, 447 del, 2394 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tgmed/wer_14
%WER 6.11 [ 3325 / 54402, 385 ins, 392 del, 2548 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tgmed_utt/wer_12
%WER 5.90 [ 3212 / 54402, 400 ins, 381 del, 2431 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tgmed_utt_offline/wer_12
%WER 6.59 [ 3587 / 54402, 416 ins, 450 del, 2721 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tgsmall/wer_11
%WER 6.92 [ 3762 / 54402, 392 ins, 505 del, 2865 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tgsmall_utt/wer_12
%WER 6.68 [ 3634 / 54402, 434 ins, 451 del, 2749 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_clean_tgsmall_utt_offline/wer_11
%WER 12.85 [ 6548 / 50948, 725 ins, 871 del, 4952 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_fglarge/wer_15
%WER 13.70 [ 6981 / 50948, 812 ins, 895 del, 5274 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_fglarge_utt/wer_15
%WER 13.18 [ 6715 / 50948, 787 ins, 841 del, 5087 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_fglarge_utt_offline/wer_15
%WER 13.36 [ 6805 / 50948, 765 ins, 924 del, 5116 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tglarge/wer_15
%WER 14.29 [ 7282 / 50948, 888 ins, 917 del, 5477 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tglarge_utt/wer_14
%WER 13.65 [ 6955 / 50948, 806 ins, 903 del, 5246 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tglarge_utt_offline/wer_15
%WER 15.64 [ 7969 / 50948, 676 ins, 1372 del, 5921 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tgmed/wer_16
%WER 16.68 [ 8497 / 50948, 771 ins, 1364 del, 6362 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tgmed_utt/wer_15
%WER 15.93 [ 8118 / 50948, 736 ins, 1286 del, 6096 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tgmed_utt_offline/wer_15
%WER 17.11 [ 8718 / 50948, 704 ins, 1547 del, 6467 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tgsmall/wer_15
%WER 18.12 [ 9232 / 50948, 751 ins, 1627 del, 6854 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tgsmall_utt/wer_15
%WER 17.38 [ 8855 / 50948, 736 ins, 1555 del, 6564 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_dev_other_tgsmall_utt_offline/wer_15
%WER 5.21 [ 2739 / 52576, 428 ins, 261 del, 2050 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_clean_fglarge_utt_offline/wer_12
%WER 5.32 [ 2795 / 52576, 402 ins, 298 del, 2095 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_clean_tglarge_utt_offline/wer_13
%WER 6.49 [ 3413 / 52576, 427 ins, 424 del, 2562 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_clean_tgmed_utt_offline/wer_12
%WER 7.18 [ 3774 / 52576, 469 ins, 477 del, 2828 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_clean_tgsmall_utt_offline/wer_11
%WER 13.35 [ 6987 / 52343, 808 ins, 925 del, 5254 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_other_fglarge_utt_offline/wer_14
%WER 13.79 [ 7219 / 52343, 847 ins, 953 del, 5419 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_other_tglarge_utt_offline/wer_13
%WER 16.08 [ 8416 / 52343, 746 ins, 1466 del, 6204 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_other_tgmed_utt_offline/wer_15
%WER 17.64 [ 9231 / 52343, 764 ins, 1662 del, 6805 sub ] exp/nnet2_online/nnet_ms_a_online/decode_pp_test_other_tgsmall_utt_offline/wer_14
# Results with nnet3 tdnn
# local/nnet3/run_tdnn.sh
# (4 epoch training on speed-perturbed data)
# num_params=19.3M
%WER 4.43 [ 2410 / 54402, 306 ins, 278 del, 1826 sub ] exp/nnet3/tdnn_sp/decode_dev_clean_fglarge/wer_13_1.0
%WER 4.63 [ 2520 / 54402, 369 ins, 259 del, 1892 sub ] exp/nnet3/tdnn_sp/decode_dev_clean_tglarge/wer_12_0.5
%WER 5.90 [ 3211 / 54402, 430 ins, 337 del, 2444 sub ] exp/nnet3/tdnn_sp/decode_dev_clean_tgmed/wer_12_0.0
%WER 6.66 [ 3622 / 54402, 450 ins, 415 del, 2757 sub ] exp/nnet3/tdnn_sp/decode_dev_clean_tgsmall/wer_12_0.0
%WER 11.62 [ 5922 / 50948, 727 ins, 741 del, 4454 sub ] exp/nnet3/tdnn_sp/decode_dev_other_fglarge/wer_14_0.5
%WER 12.19 [ 6209 / 50948, 863 ins, 682 del, 4664 sub ] exp/nnet3/tdnn_sp/decode_dev_other_tglarge/wer_14_0.0
%WER 14.52 [ 7396 / 50948, 789 ins, 1079 del, 5528 sub ] exp/nnet3/tdnn_sp/decode_dev_other_tgmed/wer_16_0.0
%WER 15.83 [ 8063 / 50948, 867 ins, 1141 del, 6055 sub ] exp/nnet3/tdnn_sp/decode_dev_other_tgsmall/wer_14_0.0
%WER 4.97 [ 2614 / 52576, 373 ins, 271 del, 1970 sub ] exp/nnet3/tdnn_sp/decode_test_clean_fglarge/wer_14_0.5
%WER 5.15 [ 2708 / 52576, 446 ins, 235 del, 2027 sub ] exp/nnet3/tdnn_sp/decode_test_clean_tglarge/wer_13_0.0
%WER 6.24 [ 3281 / 52576, 467 ins, 336 del, 2478 sub ] exp/nnet3/tdnn_sp/decode_test_clean_tgmed/wer_12_0.0
%WER 6.95 [ 3654 / 52576, 459 ins, 433 del, 2762 sub ] exp/nnet3/tdnn_sp/decode_test_clean_tgsmall/wer_13_0.0
%WER 12.14 [ 6352 / 52343, 883 ins, 649 del, 4820 sub ] exp/nnet3/tdnn_sp/decode_test_other_fglarge/wer_13_0.0
%WER 12.62 [ 6605 / 52343, 898 ins, 720 del, 4987 sub ] exp/nnet3/tdnn_sp/decode_test_other_tglarge/wer_13_0.0
%WER 15.10 [ 7904 / 52343, 874 ins, 1070 del, 5960 sub ] exp/nnet3/tdnn_sp/decode_test_other_tgmed/wer_13_0.0
%WER 16.29 [ 8528 / 52343, 828 ins, 1320 del, 6380 sub ] exp/nnet3/tdnn_sp/decode_test_other_tgsmall/wer_14_0.0
# Results with nnet3 tdnn
# local/nnet3/run_tdnn.sh
# (4 epoch training on speed-perturbed and volumn-perturbed "cleaned" data)
# num_params=19.3M, average training time=68.8s per job(on Tesla K80), real-time factor=1.23161
# for x in exp/nnet3_cleaned/tdnn_sp/decode_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 4.44 [ 2417 / 54402, 320 ins, 273 del, 1824 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_clean_fglarge/wer_13_1.0
%WER 4.69 [ 2553 / 54402, 414 ins, 209 del, 1930 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_clean_tglarge/wer_12_0.0
%WER 5.97 [ 3248 / 54402, 438 ins, 332 del, 2478 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_clean_tgmed/wer_12_0.0
%WER 6.66 [ 3625 / 54402, 476 ins, 366 del, 2783 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_clean_tgsmall/wer_11_0.0
%WER 11.62 [ 5922 / 50948, 721 ins, 716 del, 4485 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_other_fglarge/wer_15_0.5
%WER 12.20 [ 6218 / 50948, 748 ins, 776 del, 4694 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_other_tglarge/wer_14_0.5
%WER 14.65 [ 7462 / 50948, 853 ins, 985 del, 5624 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_other_tgmed/wer_15_0.0
%WER 15.90 [ 8103 / 50948, 833 ins, 1165 del, 6105 sub ] exp/nnet3_cleaned/tdnn_sp/decode_dev_other_tgsmall/wer_15_0.0
%WER 4.99 [ 2622 / 52576, 361 ins, 279 del, 1982 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_clean_fglarge/wer_13_1.0
%WER 5.20 [ 2732 / 52576, 449 ins, 230 del, 2053 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_clean_tglarge/wer_14_0.0
%WER 6.27 [ 3295 / 52576, 389 ins, 417 del, 2489 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_clean_tgmed/wer_13_0.5
%WER 7.08 [ 3724 / 52576, 489 ins, 425 del, 2810 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_clean_tgsmall/wer_13_0.0
%WER 12.07 [ 6320 / 52343, 763 ins, 761 del, 4796 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_other_fglarge/wer_13_0.5
%WER 12.52 [ 6551 / 52343, 786 ins, 825 del, 4940 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_other_tglarge/wer_13_0.5
%WER 14.78 [ 7737 / 52343, 807 ins, 1115 del, 5815 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_other_tgmed/wer_15_0.0
%WER 16.28 [ 8521 / 52343, 843 ins, 1258 del, 6420 sub ] exp/nnet3_cleaned/tdnn_sp/decode_test_other_tgsmall/wer_14_0.0
# Results with nnet3 tdnn+sMBR
# local/nnet3/run_tdnn_discriminative.sh
# a subset of the full list of results (using the acoustic model obtained at the end of the training):
%WER 4.07 [ 2214 / 54402, 313 ins, 195 del, 1706 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_fglarge_epoch4.adj/wer_14_1.0
%WER 4.20 [ 2286 / 54402, 337 ins, 206 del, 1743 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tglarge_epoch4.adj/wer_17_0.5
%WER 5.19 [ 2825 / 54402, 319 ins, 328 del, 2178 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgmed_epoch4.adj/wer_17_0.5
%WER 5.84 [ 3177 / 54402, 407 ins, 313 del, 2457 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgsmall_epoch4.adj/wer_15_0.0
%WER 11.07 [ 5641 / 50948, 745 ins, 577 del, 4319 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_fglarge_epoch4.adj/wer_20_0.5
%WER 11.43 [ 5821 / 50948, 782 ins, 603 del, 4436 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tglarge_epoch4.adj/wer_18_0.5
%WER 13.45 [ 6853 / 50948, 873 ins, 759 del, 5221 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgmed_epoch4.adj/wer_19_0.0
%WER 14.43 [ 7350 / 50948, 869 ins, 881 del, 5600 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgsmall_epoch4.adj/wer_18_0.0
%WER 4.58 [ 2409 / 52576, 370 ins, 197 del, 1842 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_fglarge_epoch4.adj/wer_14_1.0
%WER 4.75 [ 2496 / 52576, 408 ins, 200 del, 1888 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tglarge_epoch4.adj/wer_15_0.5
%WER 5.72 [ 3005 / 52576, 397 ins, 325 del, 2283 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgmed_epoch4.adj/wer_17_0.5
%WER 6.44 [ 3387 / 52576, 466 ins, 326 del, 2595 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgsmall_epoch4.adj/wer_15_0.0
%WER 11.30 [ 5913 / 52343, 708 ins, 701 del, 4504 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_fglarge_epoch4.adj/wer_16_1.0
%WER 11.71 [ 6128 / 52343, 798 ins, 633 del, 4697 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tglarge_epoch4.adj/wer_17_0.5
%WER 13.72 [ 7179 / 52343, 853 ins, 783 del, 5543 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgmed_epoch4.adj/wer_17_0.0
%WER 14.83 [ 7761 / 52343, 821 ins, 962 del, 5978 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgsmall_epoch4.adj/wer_18_0.0
# the full list of results
%WER 4.14 [ 2254 / 54402, 319 ins, 204 del, 1731 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_fglarge_epoch1.adj/wer_13_1.0
%WER 4.12 [ 2243 / 54402, 306 ins, 214 del, 1723 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_fglarge_epoch2.adj/wer_15_1.0
%WER 4.10 [ 2228 / 54402, 316 ins, 199 del, 1713 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_fglarge_epoch3.adj/wer_14_1.0
%WER 4.07 [ 2214 / 54402, 313 ins, 195 del, 1706 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_fglarge_epoch4.adj/wer_14_1.0
%WER 4.24 [ 2306 / 54402, 292 ins, 240 del, 1774 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tglarge_epoch1.adj/wer_15_1.0
%WER 4.24 [ 2307 / 54402, 357 ins, 186 del, 1764 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tglarge_epoch2.adj/wer_14_0.5
%WER 4.23 [ 2303 / 54402, 310 ins, 232 del, 1761 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tglarge_epoch3.adj/wer_16_1.0
%WER 4.20 [ 2286 / 54402, 337 ins, 206 del, 1743 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tglarge_epoch4.adj/wer_17_0.5
%WER 5.35 [ 2912 / 54402, 306 ins, 344 del, 2262 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgmed_epoch1.adj/wer_13_1.0
%WER 5.24 [ 2850 / 54402, 295 ins, 342 del, 2213 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgmed_epoch2.adj/wer_14_1.0
%WER 5.21 [ 2836 / 54402, 347 ins, 297 del, 2192 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgmed_epoch3.adj/wer_15_0.5
%WER 5.19 [ 2825 / 54402, 319 ins, 328 del, 2178 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgmed_epoch4.adj/wer_17_0.5
%WER 5.99 [ 3259 / 54402, 412 ins, 324 del, 2523 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgsmall_epoch1.adj/wer_14_0.0
%WER 5.89 [ 3204 / 54402, 416 ins, 306 del, 2482 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgsmall_epoch2.adj/wer_14_0.0
%WER 5.84 [ 3179 / 54402, 417 ins, 302 del, 2460 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgsmall_epoch3.adj/wer_14_0.0
%WER 5.84 [ 3177 / 54402, 407 ins, 313 del, 2457 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_clean_tgsmall_epoch4.adj/wer_15_0.0
%WER 10.98 [ 5593 / 50948, 849 ins, 480 del, 4264 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_fglarge_epoch1.adj/wer_16_0.0
%WER 10.98 [ 5596 / 50948, 727 ins, 600 del, 4269 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_fglarge_epoch2.adj/wer_20_0.5
%WER 10.94 [ 5576 / 50948, 752 ins, 555 del, 4269 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_fglarge_epoch3.adj/wer_18_0.5
%WER 11.07 [ 5641 / 50948, 745 ins, 577 del, 4319 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_fglarge_epoch4.adj/wer_20_0.5
%WER 11.44 [ 5826 / 50948, 764 ins, 619 del, 4443 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tglarge_epoch1.adj/wer_15_0.5
%WER 11.32 [ 5766 / 50948, 775 ins, 595 del, 4396 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tglarge_epoch2.adj/wer_16_0.5
%WER 11.28 [ 5749 / 50948, 759 ins, 615 del, 4375 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tglarge_epoch3.adj/wer_18_0.5
%WER 11.43 [ 5821 / 50948, 782 ins, 603 del, 4436 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tglarge_epoch4.adj/wer_18_0.5
%WER 13.51 [ 6885 / 50948, 838 ins, 803 del, 5244 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgmed_epoch1.adj/wer_17_0.0
%WER 13.39 [ 6824 / 50948, 887 ins, 717 del, 5220 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgmed_epoch2.adj/wer_16_0.0
%WER 13.35 [ 6803 / 50948, 894 ins, 730 del, 5179 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgmed_epoch3.adj/wer_17_0.0
%WER 13.45 [ 6853 / 50948, 873 ins, 759 del, 5221 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgmed_epoch4.adj/wer_19_0.0
%WER 14.63 [ 7452 / 50948, 831 ins, 946 del, 5675 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgsmall_epoch1.adj/wer_17_0.0
%WER 14.52 [ 7397 / 50948, 857 ins, 907 del, 5633 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgsmall_epoch2.adj/wer_17_0.0
%WER 14.40 [ 7338 / 50948, 853 ins, 901 del, 5584 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgsmall_epoch3.adj/wer_18_0.0
%WER 14.43 [ 7350 / 50948, 869 ins, 881 del, 5600 sub ] exp/nnet3/tdnn_sp_smbr/decode_dev_other_tgsmall_epoch4.adj/wer_18_0.0
%WER 4.63 [ 2432 / 52576, 349 ins, 239 del, 1844 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_fglarge_epoch1.adj/wer_14_1.0
%WER 4.56 [ 2395 / 52576, 347 ins, 223 del, 1825 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_fglarge_epoch2.adj/wer_15_1.0
%WER 4.56 [ 2397 / 52576, 361 ins, 209 del, 1827 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_fglarge_epoch3.adj/wer_14_1.0
%WER 4.58 [ 2409 / 52576, 370 ins, 197 del, 1842 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_fglarge_epoch4.adj/wer_14_1.0
%WER 4.82 [ 2535 / 52576, 406 ins, 227 del, 1902 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tglarge_epoch1.adj/wer_14_0.5
%WER 4.73 [ 2486 / 52576, 404 ins, 208 del, 1874 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tglarge_epoch2.adj/wer_14_0.5
%WER 4.73 [ 2489 / 52576, 355 ins, 250 del, 1884 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tglarge_epoch3.adj/wer_15_1.0
%WER 4.75 [ 2496 / 52576, 408 ins, 200 del, 1888 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tglarge_epoch4.adj/wer_15_0.5
%WER 5.78 [ 3040 / 52576, 412 ins, 295 del, 2333 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgmed_epoch1.adj/wer_13_0.5
%WER 5.73 [ 3015 / 52576, 394 ins, 310 del, 2311 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgmed_epoch2.adj/wer_15_0.5
%WER 5.76 [ 3027 / 52576, 404 ins, 296 del, 2327 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgmed_epoch3.adj/wer_14_0.5
%WER 5.72 [ 3005 / 52576, 397 ins, 325 del, 2283 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgmed_epoch4.adj/wer_17_0.5
%WER 6.55 [ 3443 / 52576, 454 ins, 360 del, 2629 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgsmall_epoch1.adj/wer_15_0.0
%WER 6.47 [ 3402 / 52576, 456 ins, 339 del, 2607 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgsmall_epoch2.adj/wer_15_0.0
%WER 6.45 [ 3389 / 52576, 456 ins, 335 del, 2598 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgsmall_epoch3.adj/wer_15_0.0
%WER 6.44 [ 3387 / 52576, 466 ins, 326 del, 2595 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_clean_tgsmall_epoch4.adj/wer_15_0.0
%WER 11.27 [ 5900 / 52343, 781 ins, 609 del, 4510 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_fglarge_epoch1.adj/wer_14_0.5
%WER 11.14 [ 5832 / 52343, 762 ins, 621 del, 4449 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_fglarge_epoch2.adj/wer_16_0.5
%WER 11.15 [ 5838 / 52343, 769 ins, 619 del, 4450 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_fglarge_epoch3.adj/wer_17_0.5
%WER 11.30 [ 5913 / 52343, 708 ins, 701 del, 4504 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_fglarge_epoch4.adj/wer_16_1.0
%WER 11.75 [ 6152 / 52343, 766 ins, 667 del, 4719 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tglarge_epoch1.adj/wer_15_0.5
%WER 11.59 [ 6068 / 52343, 764 ins, 654 del, 4650 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tglarge_epoch2.adj/wer_17_0.5
%WER 11.58 [ 6062 / 52343, 771 ins, 648 del, 4643 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tglarge_epoch3.adj/wer_18_0.5
%WER 11.71 [ 6128 / 52343, 798 ins, 633 del, 4697 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tglarge_epoch4.adj/wer_17_0.5
%WER 13.93 [ 7289 / 52343, 851 ins, 846 del, 5592 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgmed_epoch1.adj/wer_15_0.0
%WER 13.77 [ 7209 / 52343, 816 ins, 830 del, 5563 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgmed_epoch2.adj/wer_17_0.0
%WER 13.74 [ 7191 / 52343, 832 ins, 806 del, 5553 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgmed_epoch3.adj/wer_17_0.0
%WER 13.72 [ 7179 / 52343, 853 ins, 783 del, 5543 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgmed_epoch4.adj/wer_17_0.0
%WER 15.15 [ 7929 / 52343, 889 ins, 948 del, 6092 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgsmall_epoch1.adj/wer_14_0.0
%WER 14.93 [ 7817 / 52343, 847 ins, 958 del, 6012 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgsmall_epoch2.adj/wer_16_0.0
%WER 14.83 [ 7762 / 52343, 825 ins, 955 del, 5982 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgsmall_epoch3.adj/wer_17_0.0
%WER 14.83 [ 7761 / 52343, 821 ins, 962 del, 5978 sub ] exp/nnet3/tdnn_sp_smbr/decode_test_other_tgsmall_epoch4.adj/wer_18_0.0
# Results with nnet3 tdnn+chain model
# local/chain/run_tdnn_6z.sh
# (4 epoch training on speed-perturbed data)
# num_params=16.8M (12.7M after excluding the xent branch)
%WER 3.92 [ 2131 / 54402, 290 ins, 197 del, 1644 sub ] exp/chain/tdnn_6z_sp/decode_dev_clean_fglarge/wer_11_0.5
%WER 4.09 [ 2227 / 54402, 337 ins, 176 del, 1714 sub ] exp/chain/tdnn_6z_sp/decode_dev_clean_tglarge/wer_11_0.0
%WER 5.11 [ 2781 / 54402, 329 ins, 300 del, 2152 sub ] exp/chain/tdnn_6z_sp/decode_dev_clean_tgmed/wer_12_0.0
%WER 5.83 [ 3172 / 54402, 335 ins, 372 del, 2465 sub ] exp/chain/tdnn_6z_sp/decode_dev_clean_tgsmall/wer_12_0.0
%WER 10.43 [ 5314 / 50948, 528 ins, 697 del, 4089 sub ] exp/chain/tdnn_6z_sp/decode_dev_other_fglarge/wer_14_0.5
%WER 10.95 [ 5581 / 50948, 546 ins, 764 del, 4271 sub ] exp/chain/tdnn_6z_sp/decode_dev_other_tglarge/wer_14_0.5
%WER 13.20 [ 6723 / 50948, 676 ins, 858 del, 5189 sub ] exp/chain/tdnn_6z_sp/decode_dev_other_tgmed/wer_13_0.0
%WER 14.56 [ 7419 / 50948, 715 ins, 1003 del, 5701 sub ] exp/chain/tdnn_6z_sp/decode_dev_other_tgsmall/wer_13_0.0
%WER 4.28 [ 2251 / 52576, 292 ins, 238 del, 1721 sub ] exp/chain/tdnn_6z_sp/decode_test_clean_fglarge/wer_11_1.0
%WER 4.47 [ 2349 / 52576, 342 ins, 225 del, 1782 sub ] exp/chain/tdnn_6z_sp/decode_test_clean_tglarge/wer_11_0.5
%WER 5.55 [ 2917 / 52576, 366 ins, 314 del, 2237 sub ] exp/chain/tdnn_6z_sp/decode_test_clean_tgmed/wer_13_0.0
%WER 6.20 [ 3259 / 52576, 383 ins, 381 del, 2495 sub ] exp/chain/tdnn_6z_sp/decode_test_clean_tgsmall/wer_12_0.0
%WER 10.76 [ 5634 / 52343, 643 ins, 672 del, 4319 sub ] exp/chain/tdnn_6z_sp/decode_test_other_fglarge/wer_12_0.5
%WER 11.20 [ 5864 / 52343, 619 ins, 781 del, 4464 sub ] exp/chain/tdnn_6z_sp/decode_test_other_tglarge/wer_13_0.5
%WER 13.47 [ 7051 / 52343, 733 ins, 933 del, 5385 sub ] exp/chain/tdnn_6z_sp/decode_test_other_tgmed/wer_13_0.0
%WER 14.73 [ 7710 / 52343, 662 ins, 1209 del, 5839 sub ] exp/chain/tdnn_6z_sp/decode_test_other_tgsmall/wer_14_0.0
# Results with nnet3 tdnn+chain model
# local/chain/run_tdnn.sh
# (4 epoch training on speed-perturbed and volumn-perturbed "cleaned" data and left-biphone model)
# num_params=16.8M (12.7M after excluding the xent branch), average training time=71.8s per job(on Tesla K80), real-time factor=0.558894
# for x in exp/chain_cleaned/tdnn_sp/decode_*; do grep WER $x/wer_* | utils/best_wer.sh ; done
%WER 3.87 [ 2108 / 54402, 279 ins, 189 del, 1640 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_clean_fglarge/wer_11_0.5
%WER 3.97 [ 2162 / 54402, 277 ins, 206 del, 1679 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_clean_tglarge/wer_11_0.5
%WER 4.95 [ 2695 / 54402, 323 ins, 257 del, 2115 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_clean_tgmed/wer_11_0.0
%WER 5.57 [ 3029 / 54402, 332 ins, 307 del, 2390 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_clean_tgsmall/wer_11_0.0
%WER 10.22 [ 5208 / 50948, 547 ins, 643 del, 4018 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_other_fglarge/wer_13_0.5
%WER 10.79 [ 5497 / 50948, 568 ins, 715 del, 4214 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_other_tglarge/wer_13_0.5
%WER 13.01 [ 6628 / 50948, 718 ins, 841 del, 5069 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_other_tgmed/wer_13_0.0
%WER 14.36 [ 7317 / 50948, 700 ins, 1020 del, 5597 sub ] exp/chain_cleaned/tdnn_sp/decode_dev_other_tgsmall/wer_13_0.0
%WER 4.17 [ 2194 / 52576, 321 ins, 190 del, 1683 sub ] exp/chain_cleaned/tdnn_sp/decode_test_clean_fglarge/wer_12_0.5
%WER 4.36 [ 2292 / 52576, 331 ins, 209 del, 1752 sub ] exp/chain_cleaned/tdnn_sp/decode_test_clean_tglarge/wer_11_0.5
%WER 5.33 [ 2802 / 52576, 343 ins, 292 del, 2167 sub ] exp/chain_cleaned/tdnn_sp/decode_test_clean_tgmed/wer_11_0.5
%WER 5.93 [ 3118 / 52576, 414 ins, 287 del, 2417 sub ] exp/chain_cleaned/tdnn_sp/decode_test_clean_tgsmall/wer_10_0.0
%WER 10.62 [ 5558 / 52343, 618 ins, 690 del, 4250 sub ] exp/chain_cleaned/tdnn_sp/decode_test_other_fglarge/wer_13_0.5
%WER 10.96 [ 5738 / 52343, 628 ins, 692 del, 4418 sub ] exp/chain_cleaned/tdnn_sp/decode_test_other_tglarge/wer_12_0.5
%WER 13.24 [ 6928 / 52343, 702 ins, 895 del, 5331 sub ] exp/chain_cleaned/tdnn_sp/decode_test_other_tgmed/wer_13_0.0
%WER 14.53 [ 7603 / 52343, 726 ins, 1054 del, 5823 sub ] exp/chain_cleaned/tdnn_sp/decode_test_other_tgsmall/wer_13_0.0
# Results with nnet3 tdnn+chain+sMBR
# local/chain/run_tdnn_6z_discriminative.sh
# a subset of the full list of results (using the acoustic model obtained at the end of the training):
%WER 3.92 [ 2133 / 54402, 320 ins, 170 del, 1643 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_fglarge_epoch12/wer_11_0.5
%WER 4.03 [ 2193 / 54402, 288 ins, 213 del, 1692 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tglarge_epoch12/wer_11_1.0
%WER 5.21 [ 2835 / 54402, 306 ins, 329 del, 2200 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgmed_epoch12/wer_11_0.5
%WER 5.76 [ 3132 / 54402, 370 ins, 311 del, 2451 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgsmall_epoch12/wer_12_0.0
%WER 10.41 [ 5304 / 50948, 536 ins, 695 del, 4073 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_fglarge_epoch12/wer_14_1.0
%WER 10.94 [ 5576 / 50948, 665 ins, 650 del, 4261 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tglarge_epoch12/wer_14_0.5
%WER 13.74 [ 7002 / 50948, 733 ins, 895 del, 5374 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgmed_epoch12/wer_12_0.0
%WER 14.53 [ 7405 / 50948, 713 ins, 1033 del, 5659 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgsmall_epoch12/wer_15_0.0
%WER 4.31 [ 2268 / 52576, 326 ins, 208 del, 1734 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_fglarge_epoch12/wer_11_1.0
%WER 4.51 [ 2371 / 52576, 345 ins, 218 del, 1808 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tglarge_epoch12/wer_13_0.5
%WER 5.63 [ 2959 / 52576, 402 ins, 297 del, 2260 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgmed_epoch12/wer_11_0.0
%WER 6.15 [ 3234 / 52576, 388 ins, 365 del, 2481 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgsmall_epoch12/wer_11_0.5
%WER 10.62 [ 5558 / 52343, 668 ins, 602 del, 4288 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_fglarge_epoch12/wer_13_0.5
%WER 11.06 [ 5788 / 52343, 580 ins, 770 del, 4438 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tglarge_epoch12/wer_13_1.0
%WER 13.66 [ 7151 / 52343, 755 ins, 932 del, 5464 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgmed_epoch12/wer_12_0.0
%WER 14.75 [ 7721 / 52343, 895 ins, 841 del, 5985 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgsmall_epoch12/wer_12_0.0
# the full list of results
%WER 3.92 [ 2133 / 54402, 320 ins, 170 del, 1643 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_fglarge_epoch12/wer_11_0.5
%WER 3.91 [ 2125 / 54402, 279 ins, 217 del, 1629 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_fglarge_epoch3/wer_11_1.0
%WER 3.89 [ 2116 / 54402, 282 ins, 209 del, 1625 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_fglarge_epoch6/wer_11_1.0
%WER 3.90 [ 2122 / 54402, 281 ins, 205 del, 1636 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_fglarge_epoch9/wer_11_1.0
%WER 4.03 [ 2193 / 54402, 288 ins, 213 del, 1692 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tglarge_epoch12/wer_11_1.0
%WER 4.03 [ 2194 / 54402, 267 ins, 239 del, 1688 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tglarge_epoch3/wer_12_1.0
%WER 4.04 [ 2196 / 54402, 312 ins, 196 del, 1688 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tglarge_epoch6/wer_12_0.5
%WER 4.03 [ 2190 / 54402, 284 ins, 213 del, 1693 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tglarge_epoch9/wer_11_1.0
%WER 5.21 [ 2835 / 54402, 306 ins, 329 del, 2200 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgmed_epoch12/wer_11_0.5
%WER 5.16 [ 2808 / 54402, 295 ins, 338 del, 2175 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgmed_epoch3/wer_11_0.5
%WER 5.20 [ 2831 / 54402, 375 ins, 260 del, 2196 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgmed_epoch6/wer_10_0.0
%WER 5.23 [ 2843 / 54402, 382 ins, 256 del, 2205 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgmed_epoch9/wer_10_0.0
%WER 5.76 [ 3132 / 54402, 370 ins, 311 del, 2451 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgsmall_epoch12/wer_12_0.0
%WER 5.77 [ 3140 / 54402, 393 ins, 294 del, 2453 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgsmall_epoch3/wer_11_0.0
%WER 5.79 [ 3149 / 54402, 373 ins, 315 del, 2461 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgsmall_epoch6/wer_12_0.0
%WER 5.76 [ 3135 / 54402, 354 ins, 330 del, 2451 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_clean_tgsmall_epoch9/wer_13_0.0
%WER 10.41 [ 5304 / 50948, 536 ins, 695 del, 4073 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_fglarge_epoch12/wer_14_1.0
%WER 10.36 [ 5280 / 50948, 591 ins, 637 del, 4052 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_fglarge_epoch3/wer_15_0.5
%WER 10.42 [ 5309 / 50948, 608 ins, 622 del, 4079 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_fglarge_epoch6/wer_15_0.5
%WER 10.40 [ 5297 / 50948, 615 ins, 607 del, 4075 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_fglarge_epoch9/wer_15_0.5
%WER 10.94 [ 5576 / 50948, 665 ins, 650 del, 4261 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tglarge_epoch12/wer_14_0.5
%WER 10.91 [ 5556 / 50948, 643 ins, 667 del, 4246 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tglarge_epoch3/wer_14_0.5
%WER 10.90 [ 5551 / 50948, 646 ins, 652 del, 4253 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tglarge_epoch6/wer_14_0.5
%WER 10.92 [ 5561 / 50948, 664 ins, 644 del, 4253 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tglarge_epoch9/wer_14_0.5
%WER 13.74 [ 7002 / 50948, 733 ins, 895 del, 5374 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgmed_epoch12/wer_12_0.0
%WER 13.64 [ 6949 / 50948, 658 ins, 1027 del, 5264 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgmed_epoch3/wer_13_0.0
%WER 13.66 [ 6958 / 50948, 673 ins, 1005 del, 5280 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgmed_epoch6/wer_13_0.0
%WER 13.68 [ 6970 / 50948, 685 ins, 998 del, 5287 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgmed_epoch9/wer_13_0.0
%WER 14.53 [ 7405 / 50948, 713 ins, 1033 del, 5659 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgsmall_epoch12/wer_15_0.0
%WER 14.44 [ 7355 / 50948, 716 ins, 974 del, 5665 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgsmall_epoch3/wer_14_0.0
%WER 14.40 [ 7339 / 50948, 730 ins, 941 del, 5668 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgsmall_epoch6/wer_14_0.0
%WER 14.50 [ 7388 / 50948, 746 ins, 946 del, 5696 sub ] exp/chain/tdnn_6z_sp_smbr/decode_dev_other_tgsmall_epoch9/wer_14_0.0
%WER 4.31 [ 2268 / 52576, 326 ins, 208 del, 1734 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_fglarge_epoch12/wer_11_1.0
%WER 4.29 [ 2257 / 52576, 315 ins, 215 del, 1727 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_fglarge_epoch3/wer_11_1.0
%WER 4.30 [ 2260 / 52576, 320 ins, 213 del, 1727 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_fglarge_epoch6/wer_11_1.0
%WER 4.29 [ 2258 / 52576, 324 ins, 210 del, 1724 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_fglarge_epoch9/wer_11_1.0
%WER 4.51 [ 2371 / 52576, 345 ins, 218 del, 1808 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tglarge_epoch12/wer_13_0.5
%WER 4.48 [ 2357 / 52576, 366 ins, 203 del, 1788 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tglarge_epoch3/wer_11_0.5
%WER 4.47 [ 2349 / 52576, 363 ins, 208 del, 1778 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tglarge_epoch6/wer_12_0.5
%WER 4.50 [ 2365 / 52576, 362 ins, 208 del, 1795 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tglarge_epoch9/wer_12_0.5
%WER 5.63 [ 2959 / 52576, 402 ins, 297 del, 2260 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgmed_epoch12/wer_11_0.0
%WER 5.64 [ 2965 / 52576, 402 ins, 310 del, 2253 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgmed_epoch3/wer_11_0.0
%WER 5.61 [ 2947 / 52576, 397 ins, 296 del, 2254 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgmed_epoch6/wer_11_0.0
%WER 5.62 [ 2953 / 52576, 391 ins, 323 del, 2239 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgmed_epoch9/wer_12_0.0
%WER 6.15 [ 3234 / 52576, 388 ins, 365 del, 2481 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgsmall_epoch12/wer_11_0.5
%WER 6.10 [ 3206 / 52576, 430 ins, 316 del, 2460 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgsmall_epoch3/wer_11_0.0
%WER 6.13 [ 3223 / 52576, 422 ins, 326 del, 2475 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgsmall_epoch6/wer_12_0.0
%WER 6.15 [ 3236 / 52576, 429 ins, 322 del, 2485 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_clean_tgsmall_epoch9/wer_12_0.0
%WER 10.62 [ 5558 / 52343, 668 ins, 602 del, 4288 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_fglarge_epoch12/wer_13_0.5
%WER 10.57 [ 5532 / 52343, 659 ins, 625 del, 4248 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_fglarge_epoch3/wer_13_0.5
%WER 10.58 [ 5540 / 52343, 666 ins, 615 del, 4259 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_fglarge_epoch6/wer_13_0.5
%WER 10.64 [ 5571 / 52343, 666 ins, 609 del, 4296 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_fglarge_epoch9/wer_13_0.5
%WER 11.06 [ 5788 / 52343, 580 ins, 770 del, 4438 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tglarge_epoch12/wer_13_1.0
%WER 11.11 [ 5813 / 52343, 563 ins, 812 del, 4438 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tglarge_epoch3/wer_13_1.0
%WER 11.07 [ 5793 / 52343, 575 ins, 787 del, 4431 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tglarge_epoch6/wer_13_1.0
%WER 11.03 [ 5776 / 52343, 577 ins, 774 del, 4425 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tglarge_epoch9/wer_13_1.0
%WER 13.66 [ 7151 / 52343, 755 ins, 932 del, 5464 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgmed_epoch12/wer_12_0.0
%WER 13.60 [ 7120 / 52343, 661 ins, 1077 del, 5382 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgmed_epoch3/wer_13_0.0
%WER 13.68 [ 7158 / 52343, 737 ins, 962 del, 5459 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgmed_epoch6/wer_12_0.0
%WER 13.64 [ 7141 / 52343, 742 ins, 950 del, 5449 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgmed_epoch9/wer_12_0.0
%WER 14.75 [ 7721 / 52343, 895 ins, 841 del, 5985 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgsmall_epoch12/wer_12_0.0
%WER 14.64 [ 7664 / 52343, 818 ins, 956 del, 5890 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgsmall_epoch3/wer_13_0.0
%WER 14.70 [ 7696 / 52343, 835 ins, 945 del, 5916 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgsmall_epoch6/wer_13_0.0
%WER 14.75 [ 7722 / 52343, 892 ins, 849 del, 5981 sub ] exp/chain/tdnn_6z_sp_smbr/decode_test_other_tgsmall_epoch9/wer_12_0.0
# Results with nnet DFMSN_S + CE
# local/nnet/run_fsmn_ivector.sh DFSMN_S
# Training on the "cleaned" data
%WER 3.97 [ 2160 / 54402, 264 ins, 233 del, 1663 sub ] exp/tri7b_DFSMN_S/decode_fglarge_dev_clean/wer_10_1.0
%WER 4.10 [ 2230 / 54402, 270 ins, 242 del, 1718 sub ] exp/tri7b_DFSMN_S/decode_tglarge_dev_clean/wer_12_0.5
%WER 5.06 [ 2752 / 54402, 274 ins, 360 del, 2118 sub ] exp/tri7b_DFSMN_S/decode_tgmed_dev_clean/wer_13_0.5
%WER 5.55 [ 3022 / 54402, 335 ins, 347 del, 2340 sub ] exp/tri7b_DFSMN_S/decode_tgsmall_dev_clean/wer_11_0.0
%WER 4.61 [ 2423 / 52576, 313 ins, 288 del, 1822 sub ] exp/tri7b_DFSMN_S/decode_fglarge_test_clean/wer_12_1.0
%WER 4.69 [ 2468 / 52576, 325 ins, 282 del, 1861 sub ] exp/tri7b_DFSMN_S/decode_tglarge_test_clean/wer_13_0.5
%WER 5.69 [ 2992 / 52576, 370 ins, 344 del, 2278 sub ] exp/tri7b_DFSMN_S/decode_tgmed_test_clean/wer_11_0.5
%WER 6.28 [ 3301 / 52576, 366 ins, 428 del, 2507 sub ] exp/tri7b_DFSMN_S/decode_tgsmall_test_clean/wer_12_0.5
%WER 11.71 [ 5968 / 50948, 719 ins, 704 del, 4545 sub ] exp/tri7b_DFSMN_S/decode_fglarge_dev_other/wer_15_0.0
%WER 12.19 [ 6209 / 50948, 700 ins, 866 del, 4643 sub ] exp/tri7b_DFSMN_S/decode_tglarge_dev_other/wer_17_0.0
%WER 14.25 [ 7260 / 50948, 755 ins, 1011 del, 5494 sub ] exp/tri7b_DFSMN_S/decode_tgmed_dev_other/wer_15_0.0
%WER 15.35 [ 7823 / 50948, 741 ins, 1165 del, 5917 sub ] exp/tri7b_DFSMN_S/decode_tgsmall_dev_other/wer_15_0.0
%WER 12.01 [ 6289 / 52343, 649 ins, 930 del, 4710 sub ] exp/tri7b_DFSMN_S/decode_fglarge_test_other/wer_16_0.0
%WER 12.45 [ 6519 / 52343, 688 ins, 929 del, 4902 sub ] exp/tri7b_DFSMN_S/decode_tglarge_test_other/wer_15_0.0
%WER 14.56 [ 7622 / 52343, 715 ins, 1191 del, 5716 sub ] exp/tri7b_DFSMN_S/decode_tgmed_test_other/wer_15_0.0
%WER 15.68 [ 8208 / 52343, 743 ins, 1321 del, 6144 sub ] exp/tri7b_DFSMN_S/decode_tgsmall_test_other/wer_15_0.0
# Results with nnet DFMSN_S + CE + SMBR
# local/nnet/run_fsmn_ivector.sh DFSMN_S
# Training on the "cleaned" data
%WER 3.77 [ 2053 / 54402, 233 ins, 274 del, 1546 sub ] exp/tri7b_DFSMN_S_smbr/decode_fglarge_dev_clean/wer_27_0.5
%WER 3.86 [ 2102 / 54402, 231 ins, 264 del, 1607 sub ] exp/tri7b_DFSMN_S_smbr/decode_tglarge_dev_clean/wer_26_0.5
%WER 4.77 [ 2597 / 54402, 230 ins, 366 del, 2001 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgmed_dev_clean/wer_29_0.5
%WER 5.21 [ 2836 / 54402, 290 ins, 329 del, 2217 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgsmall_dev_clean/wer_26_0.0
%WER 4.26 [ 2239 / 52576, 250 ins, 313 del, 1676 sub ] exp/tri7b_DFSMN_S_smbr/decode_fglarge_test_clean/wer_27_1.0
%WER 4.34 [ 2282 / 52576, 265 ins, 306 del, 1711 sub ] exp/tri7b_DFSMN_S_smbr/decode_tglarge_test_clean/wer_29_0.5
%WER 5.22 [ 2746 / 52576, 319 ins, 315 del, 2112 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgmed_test_clean/wer_28_0.0
%WER 5.81 [ 3055 / 52576, 357 ins, 329 del, 2369 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgsmall_test_clean/wer_25_0.0
%WER 11.77 [ 5996 / 50948, 734 ins, 767 del, 4495 sub ] exp/tri7b_DFSMN_S_smbr/decode_fglarge_dev_other/wer_30_0.0
%WER 12.09 [ 6158 / 50948, 644 ins, 926 del, 4588 sub ] exp/tri7b_DFSMN_S_smbr/decode_tglarge_dev_other/wer_30_0.5
%WER 13.92 [ 7090 / 50948, 763 ins, 1018 del, 5309 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgmed_dev_other/wer_30_0.0
%WER 14.85 [ 7564 / 50948, 749 ins, 1108 del, 5707 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgsmall_dev_other/wer_30_0.0
%WER 11.65 [ 6099 / 52343, 595 ins, 1001 del, 4503 sub ] exp/tri7b_DFSMN_S_smbr/decode_fglarge_test_other/wer_30_0.5
%WER 12.09 [ 6329 / 52343, 686 ins, 907 del, 4736 sub ] exp/tri7b_DFSMN_S_smbr/decode_tglarge_test_other/wer_30_0.0
%WER 14.17 [ 7418 / 52343, 729 ins, 1115 del, 5574 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgmed_test_other/wer_30_0.0
%WER 15.16 [ 7933 / 52343, 759 ins, 1228 del, 5946 sub ] exp/tri7b_DFSMN_S_smbr/decode_tgsmall_test_other/wer_30_0.0
# Results with nnet DFMSN_M + CE
# local/nnet/run_fsmn_ivector.sh DFSMN_M
# Training on the "cleaned" data
%WER 4.04 [ 2200 / 54402, 254 ins, 255 del, 1691 sub ] exp/tri7b_DFSMN_M/decode_fglarge_dev_clean/wer_12_1.0
%WER 4.15 [ 2257 / 54402, 266 ins, 247 del, 1744 sub ] exp/tri7b_DFSMN_M/decode_tglarge_dev_clean/wer_13_0.5
%WER 5.01 [ 2727 / 54402, 304 ins, 308 del, 2115 sub ] exp/tri7b_DFSMN_M/decode_tgmed_dev_clean/wer_12_0.5
%WER 5.54 [ 3014 / 54402, 356 ins, 306 del, 2352 sub ] exp/tri7b_DFSMN_M/decode_tgsmall_dev_clean/wer_11_0.0
%WER 4.50 [ 2367 / 52576, 328 ins, 263 del, 1776 sub ] exp/tri7b_DFSMN_M/decode_fglarge_test_clean/wer_13_0.5
%WER 4.63 [ 2436 / 52576, 328 ins, 279 del, 1829 sub ] exp/tri7b_DFSMN_M/decode_tglarge_test_clean/wer_12_0.5
%WER 5.50 [ 2894 / 52576, 331 ins, 373 del, 2190 sub ] exp/tri7b_DFSMN_M/decode_tgmed_test_clean/wer_13_0.5
%WER 5.94 [ 3124 / 52576, 381 ins, 368 del, 2375 sub ] exp/tri7b_DFSMN_M/decode_tgsmall_test_clean/wer_13_0.0
%WER 11.88 [ 6055 / 50948, 650 ins, 891 del, 4514 sub ] exp/tri7b_DFSMN_M/decode_fglarge_dev_other/wer_17_0.5
%WER 12.24 [ 6236 / 50948, 746 ins, 811 del, 4679 sub ] exp/tri7b_DFSMN_M/decode_tglarge_dev_other/wer_17_0.0
%WER 14.18 [ 7223 / 50948, 728 ins, 1056 del, 5439 sub ] exp/tri7b_DFSMN_M/decode_tgmed_dev_other/wer_17_0.0
%WER 15.17 [ 7731 / 50948, 758 ins, 1139 del, 5834 sub ] exp/tri7b_DFSMN_M/decode_tgsmall_dev_other/wer_16_0.0
%WER 12.23 [ 6404 / 52343, 716 ins, 908 del, 4780 sub ] exp/tri7b_DFSMN_M/decode_fglarge_test_other/wer_18_0.0
%WER 12.61 [ 6598 / 52343, 736 ins, 890 del, 4972 sub ] exp/tri7b_DFSMN_M/decode_tglarge_test_other/wer_16_0.0
%WER 14.55 [ 7614 / 52343, 710 ins, 1195 del, 5709 sub ] exp/tri7b_DFSMN_M/decode_tgmed_test_other/wer_17_0.0
%WER 15.51 [ 8119 / 52343, 736 ins, 1272 del, 6111 sub ] exp/tri7b_DFSMN_M/decode_tgsmall_test_other/wer_16_0.0
# Results with nnet DFMSN_L + CE
# local/nnet/run_fsmn_ivector.sh DFSMN_L
# Training on the "cleaned" data
%WER 3.93 [ 2136 / 54402, 287 ins, 210 del, 1639 sub ] exp/tri7b_DFSMN_L/decode_fglarge_dev_clean/wer_12_0.5
%WER 3.99 [ 2170 / 54402, 279 ins, 225 del, 1666 sub ] exp/tri7b_DFSMN_L/decode_tglarge_dev_clean/wer_12_0.5
%WER 4.78 [ 2598 / 54402, 317 ins, 272 del, 2009 sub ] exp/tri7b_DFSMN_L/decode_tgmed_dev_clean/wer_12_0.0
%WER 5.20 [ 2829 / 54402, 320 ins, 303 del, 2206 sub ] exp/tri7b_DFSMN_L/decode_tgsmall_dev_clean/wer_12_0.0
%WER 4.36 [ 2294 / 52576, 285 ins, 267 del, 1742 sub ] exp/tri7b_DFSMN_L/decode_fglarge_test_clean/wer_14_0.5
%WER 4.45 [ 2342 / 52576, 295 ins, 256 del, 1791 sub ] exp/tri7b_DFSMN_L/decode_tglarge_test_clean/wer_12_0.5
%WER 5.29 [ 2782 / 52576, 320 ins, 326 del, 2136 sub ] exp/tri7b_DFSMN_L/decode_tgmed_test_clean/wer_12_0.5
%WER 5.68 [ 2988 / 52576, 372 ins, 317 del, 2299 sub ] exp/tri7b_DFSMN_L/decode_tgsmall_test_clean/wer_11_0.0
%WER 11.71 [ 5964 / 50948, 584 ins, 881 del, 4499 sub ] exp/tri7b_DFSMN_L/decode_fglarge_dev_other/wer_17_0.5
%WER 12.13 [ 6179 / 50948, 668 ins, 822 del, 4689 sub ] exp/tri7b_DFSMN_L/decode_tglarge_dev_other/wer_17_0.0
%WER 13.86 [ 7063 / 50948, 694 ins, 1051 del, 5318 sub ] exp/tri7b_DFSMN_L/decode_tgmed_dev_other/wer_17_0.0
%WER 14.82 [ 7548 / 50948, 718 ins, 1075 del, 5755 sub ] exp/tri7b_DFSMN_L/decode_tgsmall_dev_other/wer_15_0.0
%WER 12.00 [ 6281 / 52343, 724 ins, 813 del, 4744 sub ] exp/tri7b_DFSMN_L/decode_fglarge_test_other/wer_16_0.0
%WER 12.43 [ 6505 / 52343, 719 ins, 858 del, 4928 sub ] exp/tri7b_DFSMN_L/decode_tglarge_test_other/wer_16_0.0
%WER 13.99 [ 7323 / 52343, 688 ins, 1125 del, 5510 sub ] exp/tri7b_DFSMN_L/decode_tgmed_test_other/wer_17_0.0
%WER 14.90 [ 7797 / 52343, 754 ins, 1106 del, 5937 sub ] exp/tri7b_DFSMN_L/decode_tgsmall_test_other/wer_15_0.0
# Results with nnet DFMSN_L + CE + SMBR
# Training on speed-perturbed and volumn-perturbed "cleaned" data
%WER 3.60 [ 1959 / 54402, 243 ins, 171 del, 1545 sub ] exp/DFSMN_L_sp_vp_smbr/decode_fglarge_dev_clean/wer_26_0.0
%WER 3.69 [ 2010 / 54402, 194 ins, 226 del, 1590 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tglarge_dev_clean/wer_25_1.0
%WER 4.40 [ 2394 / 54402, 213 ins, 278 del, 1903 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgmed_dev_clean/wer_25_0.5
%WER 4.79 [ 2608 / 54402, 261 ins, 267 del, 2080 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgsmall_dev_clean/wer_19_0.5
%WER 3.96 [ 2083 / 52576, 280 ins, 177 del, 1626 sub ] exp/DFSMN_L_sp_vp_smbr/decode_fglarge_test_clean/wer_23_0.5
%WER 4.09 [ 2152 / 52576, 243 ins, 238 del, 1671 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tglarge_test_clean/wer_25_1.0
%WER 4.73 [ 2486 / 52576, 276 ins, 264 del, 1946 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgmed_test_clean/wer_25_0.5
%WER 5.10 [ 2682 / 52576, 304 ins, 279 del, 2099 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgsmall_test_clean/wer_21_0.5
%WER 10.21 [ 5203 / 50948, 531 ins, 650 del, 4022 sub ] exp/DFSMN_L_sp_vp_smbr/decode_fglarge_dev_other/wer_30_0.5
%WER 10.77 [ 5485 / 50948, 554 ins, 697 del, 4234 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tglarge_dev_other/wer_29_0.5
%WER 12.39 [ 6314 / 50948, 534 ins, 891 del, 4889 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgmed_dev_other/wer_29_0.5
%WER 13.01 [ 6630 / 50948, 608 ins, 885 del, 5137 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgsmall_dev_other/wer_30_0.0
%WER 10.39 [ 5439 / 52343, 560 ins, 637 del, 4242 sub ] exp/DFSMN_L_sp_vp_smbr/decode_fglarge_test_other/wer_30_0.5
%WER 10.89 [ 5702 / 52343, 580 ins, 696 del, 4426 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tglarge_test_other/wer_30_0.5
%WER 12.49 [ 6540 / 52343, 669 ins, 782 del, 5089 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgmed_test_other/wer_30_0.0
%WER 13.29 [ 6958 / 52343, 737 ins, 799 del, 5422 sub ] exp/DFSMN_L_sp_vp_smbr/decode_tgsmall_test_other/wer_26_0.0