You are on page 1of 1

Development of a Tangent Based Robust Speech

Feature Extraction Model?

Mohammad Tareq Hosain1 , Abdullah Al Arif2 , Ahmed Iqbal Pritom3 , Md


Rashedur Rahman4 , and Md. Zahidul Islam5
1
Department of CSE, Green University of Bangladesh, Dhaka, Bangladesh
mtareqhosain@gmail.com
2
Department of CSE, Green University of Bangladesh, Dhaka, Bangladesh
arif.cse.gub@gmail.com
3
Department of CSE, Green University of Bangladesh, Dhaka, Bangladesh
iqbal@cse.green.edu.bd
4
Department of CSE, Islamic University of Technology, Gazipur, Bangladesh
bishad19@iut-dhaka.edu
5
Department of CSE, Green University of Bangladesh, Dhaka, Bangladesh
zahid@cse.green.edu.bd

Abstract. An accurate speech recognition system requires close obser-


vation of the selection of an error-free speech feature extraction model.
This paper describes a prominent solution to obtain robust features from
the sound spectrum and ensures the easy recognition of speech. The pro-
posed architecture uses Tangent based (TB) auditory feature extraction
that aims to find and process robust features from the sine wave of audi-
tory signal data. This experiment suggests that every specific tune carries
distinguishing signal patterns in the spectrum diagram and hence does
the tangent of the amplitude of the same signal. To recognize the sound,
a single attribute had been used rather than using multiple attributes
where the slope of the sound spectrum being calculated.

Keywords: Tangent Based (TB) Feature Extraction · Sound Spectrum


· Signal Processing · Speech Recognition

1 Introduction
Humans have an unparalleled physical ability to be engaged in sophisticated
vocal communication. Our vocal folds, combined with the articulators, produces
a perplexing arrangement of the tune, namely “Speech”, which can be considered
as information if interpreted correctly. The speech production process includes
vent, voice and elocution [1]. Undoubtedly, speech is the most contributing factor
in linguistic messaging. It should be noted that the ingenuousness with which
humans speak is in contrast to the complication of the process.
Voice spectrogram, commonly known as voiceprint is a sophisticated way to
display speech signal which can differ by a wide margin due to different attitude
?
Supported by Green University of Bangladesh

You might also like