ASR:2014-12-22 - 版本历史

Zhangzy：/* Speaker ID */

2014-12-22T08:18:34Z

‎Speaker ID

2014年12月22日 (一) 08:05 Zhangzy

2014-12-22T08:05:49Z

Lr：以“==Speech Processing == === AM development === ==== Environment ==== * Already buy 3 760GPU * grid-9/12 760GPU crashed again; grid-11 shutdown automatically. * Chang...”为内容创建页面

2014-12-22T03:57:03Z

以“==Speech Processing == === AM development === ==== Environment ==== * Already buy 3 760GPU * grid-9/12 760GPU crashed again; grid-11 shutdown automatically. * Chang...”为内容创建页面

新页面

==Speech Processing ==
=== AM development ===

==== Environment ====
* Already buy 3 760GPU
* grid-9/12 760GPU crashed again; grid-11 shutdown automatically.
* Change 760gpu card of grid-12 and grid-14(+).
* First down-frequency of gpu760.

==== Sparse DNN ====
* Performance improvement found when pruned slightly
* need retraining for unpruned one; training loss
* details at http://liuc.cslt.org/pages/sparse.html

==== RNN AM====
* Initial nnet seems not very well, need to be pre-trained or test lower learn-rate.
* For AURORA 4 1h/epoch, model train done.
* Using AURORA 4 short-sentence with a smaller number of targets.(+)
* Adjusting the learning rate.(+)
* Trying toolkit of Microsoft.(+)
* details at http://liuc.cslt.org/pages/rnnam.html
* Reading papers

==== A new nnet training scheduler ====
* Initial code done. No better than original one considering of taking much more iterations.
* details at http://liuc.cslt.org/pages/nnet-sched.html
* Test on 4000h dataset.

====Dropout & Maxout & Convolutive network====

* Drop out(+)

:* Use different proportion of noise data to investigate the effect of xEnt and mpe and dropout
:** Problem 1) The effect of dropout in different noise proportion;
2) The effect of MPE in different noise proportion;
3) The effect of MPE+dropout in different noise proportion.

:**http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?step=view_request&cvssid=261
:* Conclusion
Dropout is effective for minority.

:** Find and test unknown noise test-data.(++)

* MaxOut
:* Pretraining based maxout, can't use large learning-rate.

* P-norm
:* Need to solve the too small learning-rate problem
:** Add one normalization layer after the pnorm-layer
:** Add L2-norm upper bound

* Convolutive network (+)
:*

====DAE(Deep Atuo-Encode)====
(1) train_clean
drop-retention/testcase(WER)| test_clean_wv1 | test_airport_wv1 | test_babble_wv1 | test_car_wv1
---------------------------------------------------------------------------------------------------------
std-xEnt-sigmoid-baseline| 6.04 | 29.91 | 27.76 | 16.37
---------------------------------------------------------------------------------------------------------
std+dae_cmvn_noFT_2-1200 | 7.10 | 15.33 | 16.58 | 9.23
---------------------------------------------------------------------------------------------------------
std+dae_cmvn_splice5_2-100 | 8.19 | 15.21 | 15.25 | 9.31
---------------------------------------------------------------------------------------------------------

:* test on XinWenLianBo music. results on
:** http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhaomy&step=view_request&cvssid=318

====Denoising & Farfield ASR====
* ICASSP paper submitted.
* HOLD

====VAD====
* Harmonics and Teager energy features being investigation (++)

====Speech rate training====
* Data ready on tencent set; some errors on speech rate dependent model. error fixed.
* Retrain new model(+)

====Scoring====
* Timber Comparison done.
* harmonics based timber comparison: frequency based feature is better. done
* GMM based timber comparison is done. Similar to speaker recognition. done
* TODO: Code checkin and '''technique report'''. done

====Confidence====
* Reproduce the experiments on fisher dataset.
* Use the fisher DNN model to decode all-wsj dataset
* preparing scoring for puqiang data
* HOLD

===Speaker ID===
* Preparing GMM-based server.
* EER ~ 4% (GMM-based system)--Text independent
* EER ~ 6%(1s) / 0.5%(5s) (GMM-based system)--Text dependent
* test different number of components; fast i-vector computing
:* Test with number recordings, The 256 number component is the best.
:* Test with text-dependent recordings, The 1024 number component is the best.
:* Interpolation alpha is not sensitive.

===Language ID===
* GMM-based language is ready.
* Delivered to Jietong
* Prepare the test-case

===Voice Conversion===
* Yiye is reading materials(+)

==Text Processing==
===LM development===

====Domain specific LM====
* domain lm
:* Sougou2T : kn-count continue .
:* lm v2.0 done,just to test the wer.

* new dict.

====tag LM====
* summary done
* need to do
:* tag Probability should test add the weight(hanzhenglong) and handover to hanzhenglong (hold)
::* paper done,begin to modify .

====RNN LM====
*rnn
:* test wer RNNLM on Chinese data from jietong-data('''this week''')
:* generate the ngram model from rnnlm and test the ppl with different size txt.[http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/Jt-chinese#sampling_data_from_rnnlm]
*lstm+rnn
:* check the lstm-rnnlm code about how to Initialize and update learning rate.(hold)

===Word2Vector===

====W2V based doc classification====

* Initial results variable Bayesian GMM obtained. Performance is not as good as the conventional GMM.(hold)
* Non-linear inter-language transform: English-Spanish-Czch: wv model training done, transform model on investigation
====Knowledge vector====
* Knowledge vector started
:* code done,to test the baseline with a task.
:* problem with weight.
====relation====
* Accomplish transE with almost the same performance as the paper did(even better)[http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=lr&step=view_request&cvssid=316:result]
====Character to word====
* Character to word conversion(hold)
:* prepare the task: word similarity
:* prepare the dict.

===Translation===

* v5.0 demo released
:* cut the dict and use new segment-tool

===QA===
====improve fuzzy match====
* add Synonyms similarity using MERT-4 method(hold)
====improve lucene search====
:* mutli query's performance improve from 66.228 to 68.672. detail:[http://cslt.riit.tsinghua.edu.cn/mediawiki/index.php/Improved_search_method_by_rewrited_query]
:* check the MERT problem that doesn't mach the qa
====XiaoI framework====
* ner from xiaoI done
====query normalization====
* using NER to normalize the word

* new inter will install SEMPRE

@@ 第80行： / 第80行： @@
 :* Non-stream GMM:wer-2.28%
     seperate3-ivector:wer-3.54 single-ivector:wer-1.57
-    seperate-PLDA:wer-0.87 single-PLDA:wer-1.04
+    seperate-PLDA:wer-0.87 single-PLDA:wer-1.00
 :* Code ready
 ===Language ID===

@@ 第4行： / 第4行： @@
 ==== Environment ====
 * Already buy 3 760GPU
 * First down-frequency of gpu760.
 ==== Sparse DNN ====
 * details at http://liuc.cslt.org/pages/sparse.html
 ==== RNN AM====
 * Adjusting the learning rate.(+)
 * Trying toolkit of Microsoft.(+)
 * details at http://liuc.cslt.org/pages/rnnam.html
 ==== A new nnet training scheduler ====
 * details at http://liuc.cslt.org/pages/nnet-sched.html
 * Test on 4000h dataset.
@@ 第30行： / 第25行： @@
 * Drop out(+)
 :**http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?step=view_request&cvssid=261
 :* Conclusion
@@ 第42行： / 第31行： @@
 :** Find and test unknown noise test-data.(++)
-* MaxOut
+* MaxOut && P-norm
-:* Pretraining based maxout, can't use large learning-rate.
-−
-* P-norm
 :* Need to solve the too small learning-rate problem
 :** Add one normalization layer after the pnorm-layer
 :** Add L2-norm upper bound
-* Convolutive network (+)
+* Convolutive network
-:*
+:* DAE test: to test various noises(car/echo/airport....)
 ====DAE(Deep Atuo-Encode)====
 :* test on XinWenLianBo music. results on
 :** http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=zhaomy&step=view_request&cvssid=318
@@ 第75行： / 第69行： @@
 ====Speech rate training====
-* Data ready on tencent set; some errors on speech rate dependent model. error fixed.
+* 64.41->34.4
-* Retrain new model(+)
-−
-====Scoring====
-* Timber Comparison done.
-* harmonics based timber comparison: frequency based feature is better. done
-* GMM based timber comparison is done. Similar to speaker recognition. done
-* TODO: Code checkin and '''technique report'''. done
 ====Confidence====
@@ 第91行： / 第78行： @@
 ===Speaker ID===
-* Preparing GMM-based server.
+:* Non-stream GMM:wer-2.28%
-* EER ~ 4% (GMM-based system)--Text independent
+   seperate3-ivector:wer-3.54 single-ivector:wer-1.57
-* EER ~ 6%(1s) / 0.5%(5s) (GMM-based system)--Text dependent
+   seperate-PLDA:wer-0.87 single-PLDA:wer-1.04
-* test different number of components; fast i-vector computing
+:* Code ready
-:* Test with number recordings, The 256 number component is the best.
-:* Test with text-dependent recordings, The 1024 number component is the best.
-:* Interpolation alpha is not sensitive.
 ===Language ID===