Sinovoice-2014-01-13 - 版本历史

2014年1月13日 (一) 07:11 Wangd

2014-01-13T07:11:38Z

Wangd：/* 6000 hour 16k training */

2014-01-13T07:10:41Z

‎6000 hour 16k training

Wangd：/* Corpora */

2014-01-13T07:09:02Z

‎Corpora

2014年1月13日 (一) 07:01 Wangd

2014-01-13T07:01:17Z

Wangd：/* 6000 hour 16k trainin */

2014-01-13T05:56:14Z

‎6000 hour 16k trainin

Wangd：/* 470 hour 8k training */

2014-01-13T05:55:16Z

‎470 hour 8k training

Wangd：以内容“=Project management= * Xiaoming and Xiao Na were added into the mail list * Potential Huawei conference-transcribing project was discussed =DNN training= ==Environme...”创建新页面

2014-01-13T05:54:26Z

以内容“=Project management= * Xiaoming and Xiao Na were added into the mail list * Potential Huawei conference-transcribing project was discussed =DNN training= ==Environme...”创建新页面

新页面

=Project management=

* Xiaoming and Xiao Na were added into the mail list
* Potential Huawei conference-transcribing project was discussed

=DNN training=

==Environment setting==

* New disk space (3T) was created and mounted at /nfs/disk1
* Jobs with 100 threads work fine on the cluster

==Corpora==
* How many extra data were obtained?

==470 hour 8k training==

* CE training done
* MPE training partially done

{| class="wikitable"
! CE !! MPE1!! MPE2 !! MPE3 !! MPE4
|-
|23.27/22.85 || 21.35/18.87 || 21.18/18.76 || 21.07/18.54
|-
|22.16/22.22 || - ||20.36/17.94 || - ||
|-
|}

==6000 hour 16k trainin==

* Audio files done. File with incorrect sampling rates were removed
* Lexicon and LM were done
* Making MFCC features

=DNN Decoder=
* Initial trail of DNN decoder based on the Sinovoice code was failed, largely due to FST compiler
* Change the strategy to an integrated approach: use the sinovoice system to control connections, and use Kaldi base for asr engine

←上一版本		2014年1月13日 (一) 07:11的版本
第41行：		第41行：
	* Initial trail of DNN decoder based on the Sinovoice code was failed, largely due to FST compiler		* Initial trail of DNN decoder based on the Sinovoice code was failed, largely due to FST compiler
	* Change the strategy to an integrated approach: use the sinovoice system to control connections, and use Kaldi base for asr engine		* Change the strategy to an integrated approach: use the sinovoice system to control connections, and use Kaldi base for asr engine
		+	* Xiaoming will do some investigation on the Sinovoice FST compiler, while Liu Chao will focus on the Kaldi-based decoder

@@ 第12行： / 第12行： @@
 ==Corpora==
-* 60 hour data were cutted
+* 60 hour data were cut this week
-* Just sending out for labeling
+* Just send out to vendors for labeling
 * Waiting for out-source platform construction
+* We assume 60 hour data per week in future
 ==470 hour 8k training==

@@ 第33行： / 第33行： @@
 ==6000 hour 16k training==
-* Audio files done. File with incorrect sampling rates were removed
+* Audio files ready. Files with incorrect sampling rates were removed
-* Lexicon and LM were done
+* Lexicon and LM were ready
 * Making MFCC features
-* Initial model can be delivered before the spring holiday
+* Initial model (6 iterations etc) can be delivered before the spring holiday
 =DNN Decoder=
 * Initial trail of DNN decoder based on the Sinovoice code was failed, largely due to FST compiler
 * Change the strategy to an integrated approach: use the sinovoice system to control connections, and use Kaldi base for asr engine

@@ 第21行： / 第21行： @@
 {| class="wikitable"
-! CE !! MPE1!! MPE2 !! MPE3 !! MPE4
+! Model !! CE !! MPE1!! MPE2 !! MPE3 !! MPE4
 |-
-|23.27/22.85 || 21.35/18.87 || 21.18/18.76 || 21.07/18.54
+|4k states||23.27/22.85 || 21.35/18.87 || 21.18/18.76 || 21.07/18.54
 |-
-|22.16/22.22 || - ||20.36/17.94 || - ||
+|8k states ||22.16/22.22 || - ||20.36/17.94 || - ||
 |-
 |}