Loading...
Please wait, while we are loading the content...
Similar Documents
Unseen noise robust speech recognition using adaptive piecewise linear transformation.
| Content Provider | CiteSeerX |
|---|---|
| Author | Chijiiwa, Keigo Suzuki, Masayuki Minematsu, Nobuaki Hirose, Keikichi |
| Abstract | SPLICE is one of the speech enhancement methods based on fea-ture conversion, which shows a high performance with a relatively small amount of calculation. After modeling noisy speech features as GMM, conversion functions are obtained for individual GMM components. The original SPLICE estimates clean feature vectors as a weighted summation of the converted versions of input vec-tors. Since the conversion functions are determined and fixed only by using training data, the effectiveness of the original SPLICE will be lower in the case of unseen noisy environments. In this paper, we propose a novel method to adapt the conversion functions to work well in unseen environments. First, to realize adaptive con-version functions, we characterize those functions using their super vectors. Then, we conduct PCA on the super vectors to reduce the number of parameters to be adapted. By representing the super vec-tors through their PCA–based base functions and weights, we imple-ment an efficient adaptation method of conversion functions, which we call Eigen–SPLICE here after. Evaluation experiments show that Eigen–SPLICE has reduced word error rate by 21.0 % relative to the conventional SPLICE, and by 24.1 % relative to EMS SPLICE in the test set B of the AURORA–2 task. |
| File Format | |
| Access Restriction | Open |
| Subject Keyword | Conversion Function Super Vector Eigen Splice Original Splice Adaptive Con-version Function Weighted Summation Individual Gmm Component Small Amount Unseen Environment Test Set Speech Enhancement Method Input Vec-tors Efficient Adaptation Method High Performance Base Function Novel Method Super Vec-tors Conventional Splice Noisy Speech Feature Converted Version Evaluation Experiment Em Splice Unseen Noisy Environment Fea-ture Conversion Word Error Rate |
| Content Type | Text |
