2014-07-25

Leftover questions

• Investigating LOUDS FST.
• CLG embedded decoder plus online compiler.
• DNN-GMM co-training

AM development

Sparse DNN

• WJS sparse DNN shows a slightly better than non-sparse cases when the network is in a large scale
• Pre-training does work for DNN training (for both 4/5/6 layers)

Noise training

• Journal paper writing on going

Multilingual ASR

• With multlingual training, performance is largely retained with most of known test sets;
• However for unknown accents, performance is not stable

Drop out & convolutional network

• Zhiyong will study drop out
• Zhiyong & Mengyuan will study convolutional network

Denoising & Farfield ASR

• Use an reverberation tool to generate a new set of datasets

• xEnt results(eval 92):

              before adaptation    after adaptation
   clean:      -                          -
   near:       19.25                    12.94
   far:        59.38                    40.46

• Lasso-based reverberation cancellation got initial clean data

VAD

• Waiting for engineering work

Scoring

• Refine the acoustic model with AMIDA database. problem solved by involving both wsj and AMIDA.

Embedded decoder

• Chatting LM release
• Train two smaller network: 500x4+600, 400x4+500: on going
• Need to upload the new client code onto git
• Build a new graph with MPE3 am and chatting LM.

LM development

Domain specific LM

h2. Domain specific LM construction

h3. TAG LM

• TAG still problematic with all-to-number tag
• check the randomness of the number tag.

h3. Chatting LM

• Building chatting lexicon
• First version released (80k lexicon)

Word2Vector

W2V based doc classification

• Initial results variable Bayesian GMM obtained. Performance is not as good as the conventional GMM.

Semantic word tree

• Version v2.0 released (filter with query log)
• Please deliver to /nfs/disk/perm/data/corpora/semanticTree (Xingchao)
• Version v3.0 under going. Further refinement with Baidu Baike hierarchy

NN LM

• Character-based NNLM (6700 chars, 7gram), 500M data training done.

• Inconsistent pattern in WER were found on Tenent test sets
• probably need to use another test set to do investigation.

• Investigate MS RNN LM training

Speaker ID

• reading materials
• prepare to run sre08

Translation

• collecting more data (Xinhua parallel text, bible, name entity) for the second version
• work into text alignment
• Will release v2.0 today