Mobile microphone robust acoustic feature identifcation using coefcient of variance
One of the most challenging techniques for speech analysis applications in mobile phones is acoustic feature extraction. The adverse environment noises, diversity of microphone specifications, and various recording software have a significant effect on the values of the extracted acous...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2021
|
| Subjects: | |
| Online Access: | http://irep.iium.edu.my/94266/9/94266_Mobile%20microphone%20robust%20acoustic%20feature%20identifcation%20using%20coefcient%20of%20variance.pdf http://irep.iium.edu.my/94266/ https://doi.org/10.1007/s10772-021-09877-1 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | One of the most challenging techniques for speech analysis applications in mobile phones is acoustic feature extraction. The adverse environment noises, diversity of microphone specifications, and various recording software have a significant effect on the values of the extracted acoustic features. In this study, we investigate the robustness of different types of acoustic features related to time-based, frequency-based, and sustained vowel using 11 different mobile recording devices. 49 recordings of subjects reciting the Rainbow Passage and 25 recordings of sustained vowel /a/ were collected. By way of synchronous recording, we analyzed and compared the extracted 253-dimensional acoustic feature vectors in order to examine how consistent the data values between the different recording devices. The variability of data values was measured using the method of coefficient of variance. Data values with low variability were identified to be from features such as the transition parameters, amplitude modulation, contrast, Chroma, mean fundamental frequency and formants. These groups of features turn out to be more reliable than others in their dependency on the recording device specifications. |
|---|