Detailed analysis of DISE algorithm and its application

In this report, the author has a detailed discussion on the Direction-Informed Speech Extraction (DISE) method based on the performance. The application of implementing the algorithm on hand phone is also proposed. DISE is a blind source separation method. With specific direction informed, it give...

Full description

Saved in:
Bibliographic Details
Main Author: Dong, Fangming
Other Authors: Andy Khong Wai Hoong
Format: Final Year Project
Language:English
Published: 2016
Subjects:
Online Access:http://hdl.handle.net/10356/68231
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:In this report, the author has a detailed discussion on the Direction-Informed Speech Extraction (DISE) method based on the performance. The application of implementing the algorithm on hand phone is also proposed. DISE is a blind source separation method. With specific direction informed, it gives similar results as fixed beam forming producing. To test the capability of reducing undesired signal, the tests on signal to noise and interference ratio (SIR), signal to distortion ratio (SDR) and signal to artifacts ratio (SAR) are performed with the variable of input SIR, microphone number, reverberation time and source number. The results show DISE has a better overall performance that fixed beamforming. The performance for sound classification is adopted from peer work. Overall it has a negative effect on sound classification. The application of the algorithm on phone is tested with 12 designs for 3 to 5 microphones. The designs with most aperture size give the best performances. To further test the algorithm in real life, more tests are suggested to be carried on with more variation of input such as different source heights, different source types and more tests that implemented on physical model is also recommended. Keywords: Blind Source Separation, DISE, Fixed Beamforming, Delay and Sum, Filter and Sum, Noise Reducing on Cell Phone, Sound Classification.