Effectiveness of multiscale fractal dimension-based phonetic segmentation in speech synthesis for low resource language

Abstract

Phonetic segmentation plays a key role in developing various speech applications. In this work, we propose to use various features for automatic phonetic segmentation task for forced Viterbi alignment and compare their effectiveness. We propose to use novel multiscale fractal dimension-based features concatenated with MelFrequency Cepstral Coefficients (MFCC). The novel features are expected to capture additional nonlinearities in speech production which should improve the performance of segmentation task. However, to evaluate effectiveness of these segmentation algorithms, we require manual accurate phoneme-level labeled data which is not available for low resource languages such as Gujarati (a low resource language and one of the official languages of India). In order to measure effectiveness of various segmentation algorithms, HMM-based speech synthesis system (HTS) for Gujarati have been built. From the subjective and objective evaluations, it is observed that FD-based features for segmentation work moderately better than other state-ofthe-art features such as MFCC, Perceptual Linear Prediction Cepstral Coefficients (PLP-CC), Cochlear Filter Cepstral Coefficients (CFCC), and RelAtive SpecTrAl(RASTA)-based PLP-CC. The Mean Opinion Score (MOS) and the Degraded-MOS, which are the measures of naturalness indicate an improvement of 9.69% with the proposed features from the MFCC (which is found to be the best among the other features) based features.

Novel Pre-processing using Outlier Removal in Voice Conversion

“…For text-dependent VC, first task is to align spectral features extracted from the source and target speakers' parallel utterances. It has been proved experimentally that alignment accuracy will impact the quality of speech in speech synthesis [3], [4] as well as in VC [5]. In the case of parallel data, Dynamic Time Warping (DTW) algorithm is used for alignment.…”

Modulation Components and Genetic Algorithm for Speaker Recognition System

“…Although this model presented some brilliant results regarding speech or speaker recognition techniques [6], [7], [8], [9]. However, its well known that some phenomena can not be captured by this model [10]. The speech instability and turbulence and other fluctuated and nonlinear open and close cycles in larynx all these phenomena can not be estimated well be the traditional source-filer model.…”