Sinovoice-2014-03-18 - 版本历史

Cslt：/* Corpora */

2014-03-21T08:09:36Z

‎Corpora

Cslt：/* Language modeling */

2014-03-19T03:26:03Z

‎Language modeling

Cslt：/* Hubei telecom */

2014-03-19T03:25:38Z

‎Hubei telecom

Cslt：/* Train Analysis */

2014-03-19T03:24:16Z

‎Train Analysis

Cslt：/* 6000 hour 16k training */

2014-03-19T03:19:50Z

‎6000 hour 16k training

Cslt：/* Corpora */

2014-03-19T03:18:58Z

‎Corpora

Cslt：以内容“=Environment setting= * Raid215 is a bit slow. Move some den-lattice and alignment to Raid212. =Corpora= * PICC data are under labeling (200h) done. * Huibei telecom...”创建新页面

2014-03-18T10:05:29Z

以内容“=Environment setting= * Raid215 is a bit slow. Move some den-lattice and alignment to Raid212. =Corpora= * PICC data are under labeling (200h) done. * Huibei telecom...”创建新页面

新页面

=Environment setting=

* Raid215 is a bit slow. Move some den-lattice and alignment to Raid212.

=Corpora=

* PICC data are under labeling (200h) done.
* Huibei telecom data labeling (108h) done.
* Now totally 1229h (470 + 346 + 105BJ mobile + 200 PICC + 108h) telephone speech is ready.
* 16k 6000h data: 978h online data from DataTang + 656h online mobile data + 4300h recording data.
* LM corpus done

=Acoustic modeling=

==Telephone model training==

===1000h Training===

* Xent completed. Compiling lattices.
* Need to test the xEnt performance

===PICC dedicated training===

* Need to collect financial text data and retrain the LM
* Need to comb word list and training text

==6000 hour 16k training==

===Training progress===

* 6000h/CSLT phone set alignment/denlattice completed
* 6000h/jt phone set alignment/denlattice completed

===Train Analysis===
* The Qihang model used a subset of the 6k data
:* 2500+950H+tang500h*+20131220, approximately 1700+2400 hours

* GMM training using this subset achieved 22.47%. Xiaoming's result is 16.1%.
:* Seems the database is not very consistent as well
:* Xiaoming will try to reproduce the Qihang training using the big database

* Test 1700h model and 6000h model on T test

<pre>
model/testcase | ditu | due1| entity1 | rec1 | shiji | zaixian1 | zaixian2 | kuaisu
------------------------------------------------------------------------------------------------
1700h_mpe | 12.18 | 12.93 | 5.29 | 3.69 | 21.73 | 25.38 | 19.45 | 12.50
------------------------------------------------------------------------------------------------
6000h_xEnt | 11.13 | 10.12 | 4.64 | 2.80 | 17.67 | 27.45 | 23.23 | 10.98
</pre>

:* 6000h data is general better than 1700h for careful reading or domain specific recording
:* 6000h with MPE/jt phoneset is on training, but better performance is expected
:* Suggest test the 6000 model on jidong
:* Suggest online test prefers online training data

===Hubei telecom===

* Hubei telecom data (127 h), retrieve 60k sentence by conf thred=0.9, amounting to 50%

<pre>
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
</pre>

* retrieve 30k sentences by conf thred=0.95, amounting to 25%, plus the original 770h data

<pre>
xEnt org: - wer_15 29.05
MPE iter1: - wer_15: 29.36
</pre>

* Incremental training with Hubei telecome based on the model (470+300+BJmobile)
:* MPE4 modeltraining done: org: 27.30, Hubei model: 25.42

=Language modeling=

* Training data ready
* Focus on PICC test set

=DNN Decoder=

==Online decoder adaptation==
* Finished alignment/den-lattice
* 1st round MPE training on going, 2 days/iteration

@@ 第7行： / 第7行： @@
 * PICC data are done (200h).
 * Huibei telecom data are done (108h).
-* Now totally 1229h (470 + 346 + 105BJ mobile + 200 PICC + 108h) telephone speech is ready.
+* Now totally 1229h (470 + 346 + 105BJ mobile + 200 PICC + 108h HBTc) telephone speech is ready.
 * 16k 6000h data: 978h online data from DataTang + 656h online mobile data + 4300h recording data.
 * LM corpus preparation done.

@@ 第67行： / 第67行： @@
 * Training data ready
-* Focus on PICC test set
+* First focus on PICC test set, try to improve the PPL
-−
 =DNN Decoder=

@@ 第60行： / 第60行： @@
 ===Hubei telecom===
-* Hubei telecom data (127 h), retrieve 60k sentence by conf thred=0.9, amounting to 50%
+* Incremental training with Hubei telecome data based on the (470+300+BJmobile) model. MPE4 finished
+:* The original model: 27.30,
-<pre>
+:* The adapted model: 25.42
-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
-</pre>
-−
-* retrieve 30k sentences by conf thred=0.95, amounting to 25%, plus the original 770h data
-−
-<pre>
-xEnt org:     -             wer_15  29.05
-MPE iter1:    -             wer_15: 29.36
-</pre>
-−
-* Incremental training with Hubei telecome based on the model (470+300+BJmobile)
-:* MPE4 modeltraining done: org: 27.30, Hubei model: 25.42
 =Language modeling=

@@ 第40行： / 第40行： @@
 * GMM training using this subset achieved 22.47%. Xiaoming's result is 16.1%.
-:* Seems the database is not very consistent as well
+:* Seems the database is still not very consistent
-:* Xiaoming will try to reproduce the Qihang training using the big database
+:* Xiaoming will try to reproduce the Qihang training using this subset
-* Test 1700h model and 6000h model on T test
+* Tested the 1700h model and 6000h model on the T test sets
 <pre>
@@ 第53行： / 第53行： @@
 </pre>
-:* 6000h data is general better than 1700h for careful reading or domain specific recording
+:* 6000h model is general better than the 1700h for careful reading or domain specific recording
-:* 6000h with MPE/jt phoneset is on training, but better performance is expected
+:* 6000h with MPE/jt phone set is still on training, but better performance is expected
-:* Suggest test the 6000 model on jidong
+:* This indicates that we should prepare domain-specific AM (not only 8k/16k). The online test prefers online training data
-:* Suggest online test prefers online training data
+:* Suggest test the 6000 model on jidong data
-−
 ===Hubei telecom===

@@ 第32行： / 第32行： @@
 * 6000h/CSLT phone set alignment/denlattice completed
 * 6000h/jt phone set alignment/denlattice completed
@@ 第79行： / 第80行： @@
 * Incremental training with Hubei telecome based on the model (470+300+BJmobile)
 :* MPE4 modeltraining done: org: 27.30, Hubei model: 25.42
 =Language modeling=

@@ 第5行： / 第5行： @@
 =Corpora=
-* PICC data are under labeling (200h) done.
+* PICC data are done (200h).
-* Huibei telecom data labeling (108h) done.
+* Huibei telecom data are done (108h).
 * Now totally 1229h (470 + 346 + 105BJ mobile + 200 PICC + 108h) telephone speech is ready.
 * 16k 6000h data: 978h online data from DataTang + 656h online mobile data + 4300h recording data.
-* LM corpus done
+* LM corpus preparation done.
 =Acoustic modeling=