Malayalam is one of the 22 scheduled languages in India with more than 130 million speakers. This paper presents a report on the development of a speaker independent, continuous transcription system for Malayalam. The system employs Hidden Markov Model (HMM) for acoustic modeling and Mel Frequency Cepstral Coefficient (MFCC) for feature extraction. It is trained with 21 male and female speakers...

Nowadays the electronic gadgets have been updated to store large amount of music information. It is necessary to have an efficient retrieval system to choose the required data. The important task in audio retrieval system is feature extraction. In the feature extraction stage, the feature which gives relevant information about music has to be extracted. In this paper, various Mel based feature ...

The process of converting an acoustic waveform into the text resembling the information, conveyed by the speaker is termed as speech recognition. Nowadays, normally Hidden Markov Model (HMM) based speech recognizer with Mel Frequency Cepstral Coefficient (MFCC) feature extraction is used. The proposed speech feature vector is generated by projecting an observed vector onto an Integrated Phoneme...

We introduce a set of speaker dependent features derived from the positions of vowels in Mel-Frequency Cepstral Coefficient (MFCC) space relative to a reference vowel. The MFCCs for a particular speaker are transformed using simple operations into features that can be used to classify vowels from a common reference point. Classification performance of vowels using Gaussian Mixture Models (GMMs)...

In this paper, we apply a discriminative weight training to a support vector machine (SVM) based gender identification. In our approach, the gender decision rule is derived by the SVM incorporating the optimally weighted mel-frequency cepstral coefficient (MFCC) based on a minimum classification error (MCE) method which is different from the previous works in that optimal weights are differentl...

This study proposes using units smaller than words, such as phonemes and syllables, as base units for speech recognition. The system presented here was developed with a hierarchical recognition logic based on the production characteristics of phonemes in Brazilian Portuguese. Decisions are made by Support Vector Machine neural networks grouped to form Specialist Machines. The descriptors used w...

In this paper, we first present a new variant of Gaussian restricted Boltzmann machine (GRBM) called multivariate Gaussian restricted Boltzmann machine (MGRBM), with its definition and learning algorithm. Then we propose using a learned GRBM or MGRBM to extract better features for robust speech recognition. Our experiments on Aurora2 show that both GRBM-extracted and MGRBM-extracted feature per...

The Mel frequency cepstral coefficient (MFCC) model, which is widely used in speech detection and recognition, is introduced to extract features from hyperspectral image data. The similarities and differences between speech signals and spectral image data are compared and analyzed. The standard MFCC model is then improved to suit the characteristics of spectral image data by reintroducing the d...

Acoustic-to-articulatory inversion of speech signals via an analysisby-synthesis method requires the comparison of natural and synthetic speech spectra either indirectly via formant frequencies, or directly via cepstral coefficients. This paper investigates several strategies of cepstral adaptation (affine transformation of cepstral coefficients, bilinear or piecewise linear frequency warping) ...

In this paper we study low-variance multi-taper spectrum estimation methods to compute the mel-frequency cepstral coefficient (MFCC) features for robust speech recognition. In speech recognition, MFCC features are usually computed from a Hamming-windowed DFT spectrum. Although windowing helps in reducing the bias of the spectrum, but variance remains high. Multitaper spectrum estimation methods...