尚未登入 請由此登入
 
 
 










知識分享平台Eshare檢視資訊View
返回前一頁Back
      [檢舉]

標題Title: An Objective Identification of Spectral Distinctiveness on Acoustic cue to Subjects with Hearing Loss
作者Authors: 李明達,..等
上傳單位Department: 電機工程系
上傳時間Date: 2013-6-7
上傳者Author: 李明達
審核單位Department: 電機工程系
審核老師Teacher: 陳培展
檔案類型Categories: 課堂報告In-class Report
關鍵詞Keyword: Acoustic cue,Spectral Distinctiveness
摘要Abstract: Hearing loss would cause serious effects on speech and language development, communication, and learning. To reduce those effects, the assistive listening devices such as hearing aids and cochlear implants could help them to use their residual hearing. Since speech perception refers to the perceptual mapping from speech signal to a linguistic representation, the speech-language pathologists provide speech-perception training to increase the subject’s ability to distinguish one phoneme from another. In clinical practice, they have to manually enhance the acoustic cues and use this information to teach a subject the differences between two phonemes, which is a time-consuming and expensive process. Therefore, it is very important to develop an automatic process to enhance the acoustic cues. To achieve this purpose, an objective identification of spectral distinctiveness on acoustic cue is proposed in this paper. In order to represent the characteristics of acoustic speech signal, the coefficients of mel-frequency cepstrum and perceptual linear prediction were selected as perceptual parameters in this study. The speech characteristics of different phonemes always vary in time, thus, the Viterbi algorithm was integrated to find the best matching condition in time domain. To estimate the differences in frequency domain, the acoustic speech signal was decomposed to the different frequency bands and the coefficients of mel-frequency cepstrum and perceptual linear prediction were also used to represent the decomposed speech signal. Finally, the spectral distinctiveness of acoustic cue in time domain and frequency domain could be identified by integrating the best matching condition and coefficients of each frequency bands. To evaluate the performance of proposed approach, the speech signal was manipulated based on identified spectral distinctiveness. Then, a close-set detection evaluation methodology was applied to evaluate the correction of spectral distinctiveness. The experimental results demonstrated that our approach is able to identify a suitable spectral distinctiveness.

檔案名稱
FileName
檔案大小
Size
檔案格式
Format
瀏覽次數
Browses
下載次數
Downloads
2013_6_688e72ae.ppt 4978Kb ppt 59 21
文件中檔案:
 

開啟檔案Download
 
 
返回前一頁