Multilingual and multimode phone recognition system for Indian languages

Abstract

The aim of this paper is to develop a flexible framework capable of automatically recognizing phonetic units present in a speech utterance of any language spoken in any mode. In this study, we considered two modes of speech: conversation, and read modes in four Indian languages, namely, Telugu, Kannada, Odia, and Bengali. The proposed approach consists of two stages: (1) Automatic speech mode classification (SMC) and(2) Automatic phonetic recognition using mode-specific multilingual phone recognition system (MPRS). In this work, vocal tract and excitation souce features are considered for speech mode classification (SMC) task. SMC systems are developed using multilayer perceptron (MLP). Further, vocal tract, excitation source, and tandem features are used to build the deep neural network (DNN)-based MPRSs. The performance of the proposed approach is compared with mode-dependent MPRSs. Experimental results show that the proposed approach which combines both SMC and MPRS into a single system outperforms the baseline mode-dependent MPRSs.

Effect of vocal tract dynamics on neural network‐based speech recognition: A Bengali language‐based study

“…We, therefore, selected a basic feedforward NN model to study vocal tract dynamics. A similar study on speech classification (Tripathi et al, 2020) also used such a basic NN model. The following subsections explain the elements involved in the classification model.…”

“…Thus, lower parameterized NN poorly performs in word classification. Tripathi et al (2020) classified continuous speech into conversational and read modes and later developed a mode-specific phone recognition system. They reported 0.83 speech mode classification accuracy with vocal tract features.…”

“…According to Tripathi et al (2020), vocal tract features achieve better speech classification than excitation source features. Although vocal tract dynamics play a significant role in speech classification, research reports, especially investigations on performance deviation due to vocal tract dynamics—either using GMM‐HMM or NN classifier—are rarely found.…”

“…Several notable works are available on speech classification; for example, Tripathi et al (2020) Analysis of articulatory feature variations has several biomedical applications as it can help diagnose several diseases (Damper, 1982). Brabenec et al (2017) utilized 14 combinations of speech tasks and acoustic features to diagnose Parkinson's disease.…”

A Comparison of Cepstral Features in the Detection of Pathological Voices by Varying the Input and Filterbank of the Cepstrum Computation

“…Among various existing cepstral features, MFCCs can be regarded as the defacto standard feature set in the area of voice pathology detection. MFCCs have also been widely used as default reference features in many areas outside pathology detection (like speaker recognition [38], speech spoof detection [39], speech mode classification [40], etc.). Moreover, MFCC features are widely included in larger generic feature sets (such as the openSMILE feature set [41], GeMAPS feature set [42], and ComParE feature set [43]) to capture vocal tract information from speech and voice signals.…”