2013-09-13 - 版本历史

Wangd：/* FBank features */

2013-09-13T04:06:47Z

‎FBank features

2013年9月13日 (五) 03:09 Wangd

2013-09-13T03:09:11Z

2013年9月13日 (五) 02:57 Wangd

2013-09-13T02:57:54Z

2013年9月13日 (五) 02:57 Wangd

2013-09-13T02:57:38Z

2013年9月13日 (五) 02:40 Wangd

2013-09-13T02:40:04Z

2013年9月13日 (五) 02:30 Wangd

2013-09-13T02:30:09Z

2013年9月13日 (五) 02:15 Wangd

2013-09-13T02:15:06Z

2013年9月13日 (五) 02:14 Wangd

2013-09-13T02:14:22Z

2013年9月13日 (五) 01:48 Wangd

2013-09-13T01:48:30Z

2013年9月13日 (五) 01:47 Wangd

2013-09-13T01:47:08Z

@@ 第25行： / 第25行： @@
 * Choose FBank 40, test various LDA output dimension. The results show LDA is still helpful, and dimension 200 is sufficient.
-[http://192.168.0.50:3000/series/?q=&action=view&series=56%2C54%2C43%2C36&chart_type=bar formance chart]
+[http://192.168.0.50:3000/series/?q=&action=view&series=56%2C54%2C43%2C36&chart_type=bar Performance chart]
 * We need to investigate non-linear discriminative approach which is simple but leads to less information lost.

@@ 第56行： / 第56行： @@
 ==DNN confidence==
-The non-acoustic lattice-based confidence is done. The phone-based accumulated confidence is done. [http://192.168.0.50:3000/series/chart?data=/json/conf_det.json Chart]
+* The non-acoustic lattice-based confidence is done. The phone-based accumulated confidence is done. [http://192.168.0.50:3000/series/chart?data=/json/conf_det.json Chart]
-It looks like the acoustic information does not contribute much to the lattice based confidence. Which means that we need a better way to combine the acoustic and the linguistic sources with models e.g., MLP.
+* It looks like the acoustic information does not contribute much to the lattice based confidence. Which means that we need a better way to combine the acoustic and the linguistic sources with models e.g., MLP.

@@ 第54行： / 第54行： @@
 [http://192.168.0.50:3000/series/?q=&action=view&series=78%2C78.0%2C78.1%2C78.2%2C78.3%2C45&chart_type=bar Performance chart]
-===DNN confidence==
+==DNN confidence==
 The non-acoustic lattice-based confidence is done. The phone-based accumulated confidence is done. [http://192.168.0.50:3000/series/chart?data=/json/conf_det.json Chart]
 It looks like the acoustic information does not contribute much to the lattice based confidence. Which means that we need a better way to combine the acoustic and the linguistic sources with models e.g., MLP.

@@ 第54行： / 第54行： @@
 [http://192.168.0.50:3000/series/?q=&action=view&series=78%2C78.0%2C78.1%2C78.2%2C78.3%2C45&chart_type=bar Performance chart]
-==Stream decoding==
+===DNN confidence==
-* The interface for server-side is done. For embedded-side is on development.
+The non-acoustic lattice-based confidence is done. The phone-based accumulated confidence is done. [http://192.168.0.50:3000/series/chart?data=/json/conf_det.json Chart]
-* Fixed a bug which prompts intermediate results when short pause encountered.
-* Fixed a CMN bug for the last segment.
+It looks like the acoustic information does not contribute much to the lattice based confidence. Which means that we need a better way to combine the acoustic and the linguistic sources with models e.g., MLP.

@@ 第44行： / 第44行： @@
 * We trained model with a random noise approach, which samples half of the training data and add 15db white noise. We hope this rand-noise learning will improve the performance of data in noise while keeping the discriminative power of the model in clean speech.
-* It is interesting to see that two factors are important in the noisy training: (1) speech should be clean (2) speech should match the training condition. The best performance is from 20db, which is not very clean and not very mismatch. This is interesting.
+* The results are largely consistent with our expectation, that the performance on noisy data were greatly improved, while the performance on clean speech is not hurted much.
 * We are looking forward to the noisy training which introduces some noises randomly in training.
 ==Stream decoding==

@@ 第25行： / 第25行： @@
 * Choose FBank 40, test various LDA output dimension. The results show LDA is still helpful, and dimension 200 is sufficient.
-[http://192.168.0.50:3000/series/?q=&action=view&series=56%2C54%2C43%2C36&chart_type=bar formance chat]
+[http://192.168.0.50:3000/series/?q=&action=view&series=56%2C54%2C43%2C36&chart_type=bar formance chart]
 * We need to investigate non-linear discriminative approach which is simple but leads to less information lost.
-−
-−
-* FB feature is much better than both MFCC and GFCC. Probably due to the less information lost without DCT.
-* In noisy environment, GFCC obtains comparable or better performance compared to FB.
-* We need to investigate how many FBs are the most appropriate.
-* Inspired by the assumption of information lost with DCT, we need to test how another transform,  LDA,  leads to the similar information lost. We need to investigate which is the suitable dimension number for the LDA. We need to investigate non-linear discriminative approach which is simple but leads to less information lost.
-* Another assumption for the better performance with FB is that the FB is more suitable for CMN. DCT accumulates a number of noisy channels and thus exhibits more uncertain. This in turn can hardly be normalized by CMN. We need to test the performance of FB and MFCC when no CMN is introduced.
 * We can also test a simple 'the same dimension DCT'. If the performance is still worse than FB, we confirm that the problem is due to noisy channel accumulation.
 * Need to investigate Gammatone filter banks. The same idea as FB, that we want to keep the information as much as possible. And it is possible to combine FB and GFB to pursue a better performance.
@@ 第46行： / 第39行： @@
 * Accumulated DNN confidence is done. The confidence values are much more reasonable.
 * Prepare MLP/DNN-based confidence integration.
 ==Noisy training ==
-Reading the table in the last section, we observe very disapointting performance reduction with noise. And we did not see too much advantage for FB and GFCC. We examine how if we introduce the noise in training. In this experiment, 15db noise are introduced in all the training data (100 hours), and the test utterances are in various noise level. Just give the performance on the test set online1. More performance is here:
+* We trained model with a random noise approach, which samples half of the training data and add 15db white noise. We hope this rand-noise learning will improve the performance of data in noise while keeping the discriminative power of the model in clean speech.
-−
-http://cslt.riit.tsinghua.edu.cn/cgi-bin/cvss/cvss_request.pl?account=wangd&step=view_request&cvssid=118
 * It is interesting to see that two factors are important in the noisy training: (1) speech should be clean (2) speech should match the training condition. The best performance is from 20db, which is not very clean and not very mismatch. This is interesting.

@@ 第21行： / 第21行： @@
 * Choose various Fbank dimension, keep LDA output dimension as 100. FB30 seems the best.
-[http://192.168.0.50:3000/series/?q=&action=view&series=36%2C34%2C29&chart_type=bar]
+[http://192.168.0.50:3000/series/?q=&action=view&series=36%2C34%2C29&chart_type=bar Performance chart]
 * Choose FBank 40, test various LDA output dimension. The results show LDA is still helpful, and dimension 200 is sufficient.
-[http://192.168.0.50:3000/series/?q=&action=view&series=56%2C54%2C43%2C36&chart_type=barperformance chat]
+[http://192.168.0.50:3000/series/?q=&action=view&series=56%2C54%2C43%2C36&chart_type=bar formance chat]

@@ 第7行： / 第7行： @@
 === Sparse DNN ===
-* Cutting 50% of the weights, and then start to run into sticky with learning rate 0.0025.
+* Cutting 50% of the weights, and then start to run into sticky with learning rate 0.0025. Continuous pruning until 1.6% weights left.
-−
 [http://192.168.0.50:3000/series/?q=&action=view&series=75%2C70%2C69%2C66%2C65%2C64%2C63%2C62%2C61%2C60%2C59&chart_type=bar performance chart]
+* The test results show not much gains with noise.
-* The comparison shows very similar performance.
+* 1/8 sparsity shows no evident performance reduction, as we observed before and is consistent the results reported by MS.
-* Cut more weights based on up-to-now sparse model. Lead to iterative sparsity.
-* Test on noisy data with the sparse.
 === FBank features ===
-Test on 100 hour data, structure 100_1200_1200_1200_1200_3580. Test on clean & 15db noiy speech.
+* CMN shows similar impact to MFCC & FBank. Since MFCC involves summary of various random channels, the mean and covariance of the dimensions are less random. This leads to two possible impacts: first, the dimensions are relatively stable therefore CMVN does not contribute much; on other hand, estimation of mean and variance is more accurate so CMVN leads to more reliable results. This means CMVN leads to unpredictable performance improvement for MFCC & Fbank, depending on the data set.
 * FB feature is much better than both MFCC and GFCC. Probably due to the less information lost without DCT.

←上一版本		2013年9月13日 (五) 01:48的版本
第10行：		第10行：


−	http://192.168.0.50:3000/series/?q=&action=view&series[]=75&~~series[]~~=~~70&series[~~]~~=69&series[]=66&series[]=65&series[]=64&series[]=63&series[]=62&series[]=61&series[]=60&series[]=59~~	+	[http://192.168.0.50:3000/series/?q=&action=view&series=75%2C70%2C69%2C66%2C65%2C64%2C63%2C62%2C61%2C60%2C59&chart_type=bar performance chart]
−		+

←上一版本		2013年9月13日 (五) 01:47的版本
第10行：		第10行：


−	[http://192.168.0.50:3000/series/?q=&action=view&series[]=75&series[]=70&series[]=69&series[]=66&series[]=65&series[]=64&series[]=63&series[]=62&series[]=61&series[]=60&series[]=59~~\|Performance chart]~~	+	http://192.168.0.50:3000/series/?q=&action=view&series[]=75&series[]=70&series[]=69&series[]=66&series[]=65&series[]=64&series[]=63&series[]=62&series[]=61&series[]=60&series[]=59