Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization
碩士 === 國立臺灣大學 === 電信工程學研究所 === 100 === In this paper we propose a new approach of two-dimensional frame-and-feature weighted Viterbi decoding performed at the recognizer back-end for robust speech recognition. The frame weighting is based on an Support Vector Machine (SVM) classifier considering the...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2012
|
Online Access: | http://ndltd.ncl.edu.tw/handle/80127356350852988068 |
id |
ndltd-TW-100NTU05435053 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-100NTU054350532015-10-13T21:50:16Z http://ndltd.ncl.edu.tw/handle/80127356350852988068 Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization 使用二維特徵音框權重法及調變頻譜正規化之強健型語音辨識 Yang Chang 張暘 碩士 國立臺灣大學 電信工程學研究所 100 In this paper we propose a new approach of two-dimensional frame-and-feature weighted Viterbi decoding performed at the recognizer back-end for robust speech recognition. The frame weighting is based on an Support Vector Machine (SVM) classifier considering the energy distribution and cross-correlation spectrum of the frame. The basic idea is that voiced frames with higher harmonicity is in general more reliable than other frames in noisy speech and therefore should be weighted higher. The feature weighting is based on an entropy measure considering confusion between phoneme classes. The basic idea is that the scores obtained with more discriminating features causing less confusion between phonemes should be weighted higher. These two different weighting schemes on the two different dimensions, frames and features, are then properly integrated in Viterbi decoding. Very significant improvements were achieved in extensive experiments performed with the Aurora 4 testing environment for all types of noise and all SNR values. 李琳山 2012 學位論文 ; thesis 87 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立臺灣大學 === 電信工程學研究所 === 100 === In this paper we propose a new approach of two-dimensional frame-and-feature weighted Viterbi decoding performed at the recognizer back-end for robust speech recognition. The frame weighting is based on an Support Vector Machine (SVM) classifier considering the energy distribution and cross-correlation spectrum of the frame. The basic idea is that voiced frames with higher harmonicity is in general more reliable than other frames in noisy speech and therefore should be weighted higher. The feature weighting is based on an entropy measure considering confusion between phoneme classes. The basic idea is that the scores obtained with more discriminating features causing less confusion between phonemes should be weighted higher. These two different weighting schemes on the two different dimensions, frames and features, are then properly integrated in Viterbi decoding. Very significant improvements were achieved in extensive experiments performed with the Aurora 4 testing environment for all types of noise and all SNR values.
|
author2 |
李琳山 |
author_facet |
李琳山 Yang Chang 張暘 |
author |
Yang Chang 張暘 |
spellingShingle |
Yang Chang 張暘 Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization |
author_sort |
Yang Chang |
title |
Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization |
title_short |
Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization |
title_full |
Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization |
title_fullStr |
Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization |
title_full_unstemmed |
Robust Speech Recognition with Two-dimensional Frame-and-feature Weighting and Modulation Spectrum Normalization |
title_sort |
robust speech recognition with two-dimensional frame-and-feature weighting and modulation spectrum normalization |
publishDate |
2012 |
url |
http://ndltd.ncl.edu.tw/handle/80127356350852988068 |
work_keys_str_mv |
AT yangchang robustspeechrecognitionwithtwodimensionalframeandfeatureweightingandmodulationspectrumnormalization AT zhāngyáng robustspeechrecognitionwithtwodimensionalframeandfeatureweightingandmodulationspectrumnormalization AT yangchang shǐyòngèrwéitèzhēngyīnkuāngquánzhòngfǎjídiàobiànpínpǔzhèngguīhuàzhīqiángjiànxíngyǔyīnbiànshí AT zhāngyáng shǐyòngèrwéitèzhēngyīnkuāngquánzhòngfǎjídiàobiànpínpǔzhèngguīhuàzhīqiángjiànxíngyǔyīnbiànshí |
_version_ |
1718068542126948352 |