The Research of Speech Enhancement Based on Reverberation Detection

• Main Authors: Hung-wei Hsieh, 謝宏暐
• Other Authors: Shyang-Jye Chang
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/97885697842841212005

碩士 === 國立雲林科技大學 === 機械工程系碩士班 === 100 === The purpose of this experiment is to tone up the design signal by a noise, which the sound make up by undesired sounds, reverberated sounds, and a design signal. For instance, when we are meeting in a reverberated space, using a microphone in some ones’ bedroom, or using a laptop without using external microphone, even talking to a mobile phone far away from you, all of these conditions, we do not want any undesired sounds produced. In this research, we use the sound velocity definition and the ‘reverberation’, to find out the second high wave crest, the design signal reflected from the reverberated face at a few seconds later, then let the wave signal combine with the design signal, which we wanted in a noise signal, to make the design signal more powerful rather than other undesired sounds. The findings revealed that when the microphone using at the specific position have the best experimental results, and we add the ‘Time Shifting’ method to compare with the method we defined with ‘convolution’, to know when and what method we chosen is good to make a clear desired sound. Base on the findings, if we use the method to heighten the sound what we need, we can heighten the recording range, and decrease more mathematical counts.