“Sinovoice-2014-03-11”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
Training Analysis
 
(相同用户的2个中间修订版本未显示)
第18行: 第18行:
 
* Expect to finish in 7 days
 
* Expect to finish in 7 days
  
===BJ mobile incremental training===
 
 
(1) Original 470 + 300 model: 30.24% WER
 
<pre>
 
MPE2      MPE3        MPE3+iLM      MPE4+iLM
 
27.01%    26.72%      25.09%        24.53%
 
</pre>
 
  
 
===PICC dedicated training===
 
===PICC dedicated training===
第30行: 第23行:
 
<pre>
 
<pre>
 
Baseline (470+300h): 45.03
 
Baseline (470+300h): 45.03
+ PICC 105h incremental training (th=0.9): 41.89
+
+ PICC 188h incremental training (th=0.9): 41.89
+ PICC 105h incremental training (th=0.8): 41.64
+
+ PICC 188h incremental training (th=0.8): 41.64
+ PICC 105h labelled training: 34.78
+
+ PICC 188h labelled training: 34.78
+ PICC 105h labelled training + PICC text LM: 29.18
+
+ PICC 188h labelled training + PICC text LM: 29.18
 
</pre>
 
</pre>
 
  
 
==6000 hour 16k training==
 
==6000 hour 16k training==
第67行: 第59行:
 
* HTK training on the same database
 
* HTK training on the same database
 
:* HLDA:    18.22
 
:* HLDA:    18.22
:* HLDA+MPE: 14.40
+
:* HLDA+MPE: 17.40
 
+
  
 
===Hubei telecom===
 
===Hubei telecom===

2014年3月11日 (二) 06:42的最后版本

Environment setting

  • Raid212/Raid215/Disk212 done

Corpora

  • PICC data are under labeling (200h) done.
  • Now totally 1121h (470 + 346 + 105BJ mobile + 200 PICC) telephone speech is ready.
  • 16k 6000h data: 978h online data from DataTang + 656h online mobile data + 4300h recording data.
  • LM training text need be prepared in 2 days.

Acoustic modeling

Telephone model training

1000h Training

  • Training recipe prepared
  • Expect to finish in 7 days


PICC dedicated training

Baseline (470+300h): 45.03
+ PICC 188h incremental training (th=0.9): 41.89
+ PICC 188h incremental training (th=0.8): 41.64
+ PICC 188h labelled training: 34.78
+ PICC 188h labelled training + PICC text LM: 29.18

6000 hour 16k training

Training progress

  • Ran DNN MPE to iteration 5.
  • Receipe
  • 100h MPE training
  • 1700h MPE alignment/lattice
  • 1700h MPE training
  • 1 week to complete 3 MPE iterations
  • MPE2 result: 1e-9: 10.67% (8.61%), 1e-10: 10.34% (8.27%)
  • MPE3 result: 1e-9: 10.48% (8.43%), 1e-10: 10.12% (8.05%)
  • MPE4 result: 1e-9: 10.34% (8.31%), 1e-10: 10.03% (7.97%)
  • MPE5 result:

Training Analysis

  • Shared tree GMM model training completed, WER% is similar to non-shared model .
  • Selected 100h online data, trained two systems: (1) di-syllable system (2) jt-phone system
        di-syl      jt-ph
GMM:      -         20.86%
Xent    15.42%      14.78%       
MPE1    14.46%      14.23%
MPE2    14.22%      14.09%
MPE3    14.26%      13.80%
MPE4    14.24%      13.68%
  • HTK training on the same database
  • HLDA: 18.22
  • HLDA+MPE: 17.40

Hubei telecom

  • Hubei telecom data (127 h), retrieve 60k sentence by conf thred=0.9, amounting to 50%
xEnt org:  -             wer_15  29.05
MPE iter1:wer_14 29.23;wer_15 29.38
MPE iter2:wer_14 29.05;wer_15 29.11
MPE iter3:wer_14 29.32;wer_15 29.28
MPE iter4:wer_14 29.29;wer_15 29.28
  • retrieve 30k sentences by conf thred=0.95, amounting to 25%, plus the original 770h data
xEnt org:     -             wer_15  29.05
MPE iter1:    -             wer_15: 29.36

Language modeling

  • Need transfer the training text

DNN Decoder

Online decoder

  • CMN code delivered. Integration is done
  • CMN pipe code delivered. Model adaptation is on going