Automatic Speech Recognition and Translation for Low Resource Languages

4Language Identification Using Speech Denoising Techniques: A Review

Amal Kumar*, Piyush Kumar Singh and Jainath Yadav

Department of Computer Science, Central University of South Bihar, Gaya, India

Abstract

Speech denoising and language identification are two important tasks in speech processing. Speech denoising is the process of removing unwanted noise from a speech signal, while language identification involves identifying the language spoken in a speech signal. In this chapter, we discuss various approaches for speech denoising of language identification. These approaches consist of two major steps. First step deals with the noisy speech signal and how it is denoised using temporal and spectral processing. For the next step, denoised signal is classified for identifying the language spoken using recent machine learning algorithm. Experimental results on a dataset of noisy speech signals in multiple languages show that the proposed approach achieves high denoising and language identification performance. The denoising step significantly improves the accuracy of language identification, as the noise can often interfere with the speech signal’s acoustic features that have a role in identifying the language. Overall, the proposed approach can be useful in various applications where accurate language identification is required, such as speech recognition, language translation, and speaker recognition systems.

Keywords: Speech denoising, language identification, filtering, denoised ...