A Study on Voice Activation Detection by Using Neural Networks
碩士 === 國立臺北大學 === 通訊工程研究所 === 105 === In this thesis, a new approach of Voice Activation Detection to considered the effects of ASR system and discussed the method of NN-based VAD. This method improve the erroneous judgment caused by the sudden gitter in speech signal when we used VAD of Energy-base...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2017
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/fk4332 |
| Summary: | 碩士 === 國立臺北大學 === 通訊工程研究所 === 105 === In this thesis, a new approach of Voice Activation Detection to considered the effects of ASR system and discussed the method of NN-based VAD. This method improve the erroneous judgment caused by the sudden gitter in speech signal when we used VAD of Energy-based and GMM-based. National Taipei University (NTPU) Additive Noise corpus of this study was established based on TCC300 corpus and smart phone noise data-based. In the outside test, multi-condition results are better than matched-condition, so it proved the multi-condition of hidden layer can learn each other.
At last, we proposed “A study on Voice Activation Detection by Using Neural Networks” what is prerequisite study of ASR system.
|
|---|