Automatic speech recognition using Mel- frequency cepstrum coefficient (MFCC) and vector quantization (VQ) techniques for continuous speech

Et Al Verma

doi:10.21833/ijaas.2018.04.009

Abstract

Automatic speech recognition using Mel- frequency cepstrum coefficient (MFCC) and vector quantization (VQ) techniques for continuous speech

Highlights

Automatic Speech recognition consists of two parts
Proposed method and implementation In Automatic Speech Recognition, speech inputs have been given with microphone and stored in a database for training purpose
Speech signal is processed by Mel- Frequency Cepstrum Coefficient (MFCC) and Vector Quantization (VQ) techniques which extract features

Summary

Introduction

Automatic Speech recognition consists of two parts. The first part is training part where the whole speaker database is created and another part is testing part where speaker recognition occurs. Different phases of speech recognition are: Feature Extraction (Speech analyzer) Matching process (Speech Classifier)

Feature extraction process

Speech matching process

VQ matching technique

Proposed method and implementation

Result and discussion

Conclusion