View
91
Download
1
Category
Preview:
DESCRIPTION
專題研究 (3) Viterbi Decoding Triphone Acoustic Model. Prof. Lin-Shan Lee, TA. Yun-Chiao Li. Viterbi Decoding. 03.04.mono0a.viterbi.sh 04.04.tri1.viterbi.sh. Viterbi Decoding. Instead of using WFST, we use Viterbi now Converted Kaldi Acoustic model to HTK by Vulcan (02.02.convert.htk.feat.sh). - PowerPoint PPT Presentation
Citation preview
專題研究 (3)Viterbi DecodingTriphone Acoustic Model
Prof. Lin-Shan Lee, TA. Yun-Chiao Li
1
03.04.mono0a.viterbi.sh04.04.tri1.viterbi.sh
Viterbi Decoding2
Viterbi Decoding
Instead of using WFST, we use Viterbi now
Converted Kaldi Acoustic model to HTK by Vulcan
(02.02.convert.htk.feat.sh)
3
Convert the acoustic model from Kaldi to
HTK
Viterbi Decoding4
Using the dev set to find the best acoustic
weight (acwt)
04.01~04.04
Triphone Acoustic Model5
Triphone Acoustic Model 6
In monophone acoustic model, ㄅ、ㄆ、ㄇ they use their own model
In triphone acoustic model, ㄅ - ㄆ - ㄇ is a model
There will be too many model and lack of training data
Decision Tree
Use decision tree to tie similar models together
7
04.01.tri1.train.sh (1/3)
It is very similar to 03.01
8
04.01.tri1.train.sh (2/3)9
04.01.tri1.train.sh (3/3)10
bash 04.01.tri1.train.shbash 04.02.tri1.mkgraph.shbash 04.03.tri1.fst.shbash 04.04.tri1.viterbi.sh
Homework11
Some Helpful References
“使用加權有限狀態轉換器的基於混合詞與次詞
以文字及語音指令偵測口語詞彙” – 第三章 https://www.dropbox.com/s/
dsaqh6xa9dp3dzw/wfst_thesis.pdf
Check HDecode, HLRescore in HTK Book
12
Recommended