|
Text based Speaker Identification Using K-means-clustering and SVM Classification
|
Hossain Khaleghi biaki   , Mostafa Kavanlooie |
|
|
|
Abstract: (2372 Views) |
| The purpose of this paper is to develop a text-based speakerIdentification system by providing a new method for speaker classification. In this work, a speaker identification system has been successfully tested for a 10-person database provided by the authors in the normal room environment with a standard headset microphone.The Mel Frequency Cepstral Coefficient (MFCC) is used as a discriminating feature which is extracted from the speech of individual persons and stored with the first and second derivatives as a property vector.Using Vector Quantization (VQ), the data has been reduced and the SupportVector Machine (SVM) with the radial base functions and polynomial core to classify the speakers.The simulation results show that the proposed algorithm has acceptable detection capability in high Signal to Noise Ratio (SNR), but detection percentage decreases by decreasing the signal-to-noise ratio. |
|
| Keywords: Mel Frequency Cepstral Coefficient, backup vector machine, SupportVector machine, Vector Quantization, Text based Speaker Identification. |
|
|
Full-Text [PDF 5451 kb]
(1717 Downloads)
|
Type of Study: Scientific-extension |
Subject:
Special
Received: 2018/07/23 | Accepted: 2018/09/23 | Published: 2018/12/21
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
