Sinovoice-2014-04-08 - 版本历史

2014年4月9日 (三) 04:34 Cslt

2014-04-09T04:34:17Z

Cslt：/* 6000 hour 16k training */

2014-04-08T09:09:46Z

‎6000 hour 16k training

Cslt：/* Environment setting */

2014-04-08T08:35:08Z

‎Environment setting

Cslt：以内容“=Environment setting= * Sinovoice internal server deployment. Gitlab + Trac. Some problems remain, need to investigate other tools. =Corpora= * 300h Guangxi telecom...”创建新页面

2014-04-08T07:28:10Z

以内容“=Environment setting= * Sinovoice internal server deployment. Gitlab + Trac. Some problems remain, need to investigate other tools. =Corpora= * 300h Guangxi telecom...”创建新页面

新页面

=Environment setting=

* Sinovoice internal server deployment. Gitlab + Trac. Some problems remain, need to investigate other tools.

=Corpora=

* 300h Guangxi telecom text transcription prepared. 150h before 18th, April.
* Now totally 1338h (470 + 346 + 105BJ mobile + 200 PICC + 108h HBTc + 109h New BJ mobile) telephone speech is ready.
* 16k 6000h data: 978h online data from DataTang + 656h online mobile data + 4300h recording data.
* Xiaona will prepare noise database. Start from telephone speech.

=Acoustic modeling=

==Telephone model training==

===1000h Training===

*Baseline: 8k states, 470+300 MPE4, 20.29
* Jietong phone, 200 hour seed, 10k states training:
:* Xent 5 iteration: 23.26
:* Xent 15 iteration done.

* CSLT phone, 8k states training
:* MPE1: 20.60
:* MPE2: 20.37
:* MPE3: 20.37
:* MPE4: 20.37

==6000 hour 16k training==

===Training progress===

* 6000h/CSLT phone set alignment/denlattice completed
:* Xent: 12.83
:* MPE1: 9.21
:* MPE2: 9.13

* 6000h/jt phone set alignment/denlattice completed
:* lattice done, MPE run 4 days

===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 still not very consistent
:* Xiaoming kicked off the job to reproduce the Qihang training using this subset

===Multilanguage Training===

* Prepare Chinglish data: Wang Dong provide info.
* Prepare shared DNN structure for multilingual training

===Noise robust feature===

* GFbank can be propagated to Sinovoice
* Liuchao will prepare fast computing code

=Language modeling=

* Training process was delivered.
* Some problems in encoding.
* Training text retrieval based on topic models (initial key words)

=DNN Decoder=

==decoder optimization==

* Test computation cost of each step
:* beam 9/5000: netforward 65%
:* beam 13/7000: netforward 28%
* Liuchao will verify the proportion number with CSLT engine.
* Zhiyong & Liuchao will deliver the frame-skipping approach.
* Investigate BigLM retrieval optimization.

@@ 第20行： / 第20行： @@
 * Jietong phone, 200 hour seed, 10k states training:
 :* Xent 5 iteration: 23.26
-:* Xent 15 iteration done.
+:* Xent 16 iteration: 22.90

@@ 第33行： / 第33行： @@
 ===Training progress===
-* 6000h/CSLT phone set alignment/denlattice completed
+* 6000h/CSLT phone set training
 :* Xent: 12.83
 :* MPE1: 9.21
 :* MPE2: 9.13
-* 6000h/jt phone set alignment/denlattice completed
+* 6000h/jt phone set  phone set training
 :* lattice done, MPE run 4 days

@@ 第1行： / 第1行： @@
 =Environment setting=
-* Sinovoice internal server deployment. Gitlab + Trac. Some problems remain, need to investigate other tools.
+* Sinovoice internal server deployment. The current configuration involves Gitlab + Trac.
+* Some access permission problems remain, need to investigate other tools.
 =Corpora=