Hi everybody,
I am sure somebody used the Adaptation method to train the new acoustic model,is there any one can show me the parameter list of "bw"?
I want to train the model base on RM1_cd_semi which comes from the sphinx3 offical release,and the parameter I used is:
bin/bw -moddeffn RM1_cd_semi/RM1.1000.mdef -ts2cbfn .semi. -mixwfn RM1_cd_semi/means -tmatfn tmatn -meanfn mean -varfn var -dictfn etc/new.dic -fdictfn etc/new.filler -ctlfn etc/new.fileids -part 1 -npart 1 -cepdir feat/ -cepext mfc -lsnfn etc/new.transcription -accumdir accumdir -meanreest yes -varreest yes -tmatreest yes -feat s2_4x
finanlly I got this:
INFO: main.c(253): Reading RM1_cd_semi/RM1.1000.mdef
INFO: model_def_io.c(587): Model definition info:
INFO: model_def_io.c(588): 39619 total models defined (49 base, 39570 tri)
INFO: model_def_io.c(589): 237714 total states
INFO: model_def_io.c(590): 1245 total tied states
INFO: model_def_io.c(591): 245 total tied CI states
INFO: model_def_io.c(592): 49 total tied transition matrices
INFO: model_def_io.c(593): 6 max state/model
INFO: model_def_io.c(594): 6 min state/model
bw: s3io.c:431: s3read_3d: Assertion `n == l_d1 * l_d2 * l_d3' failed.
I think maybe there is something wrong about parameter -ts2cbfn,I also used .cont.,it shows the same information.
and can you tell me the difference between semi continues and continues model?
Thanks a lot.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,it is me again :(
while I was running bw,I always got this message
ERROR: "backward.c", line 401: final state not reached.ERROR: "baum_welch.c", line 331: cy1_clstk/audio8618_9891-1 ignored
Is that normal?How to fix it?
Sure, just run 'bw' without any arguments and it will show you its parameter list.
The problem you are having is that you are starting from the wrong model. The RM1.cd_semi model is the untied context-dependent model. You should use RM1.cd_semi_1000 instead.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
Thank you very much.
But I try it again,and I use rm1.cd_semi_1000 model,it gave me the same information.And I know the parameter list,but I do not know exactly which parameter I should use.can you show me which parameter I 'must' provide,acording to this model?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi everybody,
I am sure somebody used the Adaptation method to train the new acoustic model,is there any one can show me the parameter list of "bw"?
I want to train the model base on RM1_cd_semi which comes from the sphinx3 offical release,and the parameter I used is:
bin/bw -moddeffn RM1_cd_semi/RM1.1000.mdef -ts2cbfn .semi. -mixwfn RM1_cd_semi/means -tmatfn tmatn -meanfn mean -varfn var -dictfn etc/new.dic -fdictfn etc/new.filler -ctlfn etc/new.fileids -part 1 -npart 1 -cepdir feat/ -cepext mfc -lsnfn etc/new.transcription -accumdir accumdir -meanreest yes -varreest yes -tmatreest yes -feat s2_4x
finanlly I got this:
INFO: main.c(253): Reading RM1_cd_semi/RM1.1000.mdef
INFO: model_def_io.c(587): Model definition info:
INFO: model_def_io.c(588): 39619 total models defined (49 base, 39570 tri)
INFO: model_def_io.c(589): 237714 total states
INFO: model_def_io.c(590): 1245 total tied states
INFO: model_def_io.c(591): 245 total tied CI states
INFO: model_def_io.c(592): 49 total tied transition matrices
INFO: model_def_io.c(593): 6 max state/model
INFO: model_def_io.c(594): 6 min state/model
bw: s3io.c:431: s3read_3d: Assertion `n == l_d1 * l_d2 * l_d3' failed.
I think maybe there is something wrong about parameter -ts2cbfn,I also used .cont.,it shows the same information.
and can you tell me the difference between semi continues and continues model?
Thanks a lot.
Hi,it is me again :(
while I was running bw,I always got this message
ERROR: "backward.c", line 401: final state not reached.ERROR: "baum_welch.c", line 331: cy1_clstk/audio8618_9891-1 ignored
Is that normal?How to fix it?
and finally I got:
WARNING: "accum.c", line 626: The following seno never occur in the input data
25 26 27 28 29 30 31 32 33 34
55 56 57 58 59 65 66 67 68 69
105 106 107 108 109 125 126 127 128 129
165 166 167 168 169 195 196 197 198 199
205 206 207 208 209 269 284 292 295 301
307 370 371 372 373 374 375 376 377 378
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
439 440 441 442 443 444 445 446 507 508
509 510 511 512 513 514 515 516 517 535
536 537 538 539 540 541 542 543 544 588
635 637 688 689 690 691 692 693 694 695
696 697 698 699 700 701 702 703 704 705
706 707 708 709 710 784 785 786 787 788
789 790 791 792 878 923 924 925 926 927
929 945 956 957 958 959 960 1100 1101 1102
1103 1104 1105 1106 1107 1108 1109 1110 1111 1112
1113 1114 1115 1116 1117 1118 1119 1120 1121 1122
1123 1124 1125 1134 1135 1136 1137 1138 1139 1140
1141 1142 1143 1210 1240 1241 1242 1243 1244
Does that means I should prepare more audio data? to cover all the seno?
> ERROR: "backward.c", line 401: final state not reached.ERROR: "baum_welch.c", line 331: cy1_clstk/audio8618_9891-1 ignored
It means your transcription probably doesn't match actual recording. Check audio8618_9891-1 once again.
> WARNING: "accum.c", line 626:
It's a warning, but of course it's nice to have more data.
Sure, just run 'bw' without any arguments and it will show you its parameter list.
The problem you are having is that you are starting from the wrong model. The RM1.cd_semi model is the untied context-dependent model. You should use RM1.cd_semi_1000 instead.
Hi,
Thank you very much.
But I try it again,and I use rm1.cd_semi_1000 model,it gave me the same information.And I know the parameter list,but I do not know exactly which parameter I should use.can you show me which parameter I 'must' provide,acording to this model?
ok,I got it ,I use the wrong parameter,the path of var mean and tmat is wrong,sorry.