No abstract

Change in voice quality (VQ) is one of the first precursors of Parkinson's disease (PD). Specifically, impacted phonation and articulation causes the patient to have a breathy, husky-semiwhisper and hoarse voice.A goal of this paper is to characterize a VQ spectrum -the composition of non-modal phonations -of voice in PD. The paper relates non-modal healthy phonations: breathy, creaky, tense, falsetto and harsh, with disordered phonation in PD. First, statistics are learned to differentiate the modal and non-modal phonations. Statistics are computed using phonological posteriors, the probabilities of phonological features inferred from the speech signal using a deep learning approach. Second, statistics of disordered speech are learned from PD speech data comprising 50 patients and 50 healthy controls. Third, Euclidean distance is used to calculate similarity of non-modal and disordered statistics, and the inverse of the distances is used to obtain the composition of non-modal phonation in PD. Thus, pathological voice quality is characterised using healthy non-modal voice quality "base/eigenspace". The obtained results are interpreted as the voice of an average patient with PD and can be characterised by the voice quality spectrum composed of 30% breathy voice, 23% creaky voice, 20% tense voice, 15% falsetto voice and 12% harsh voice. In addition, the proposed features were applied for prediction * Corresponding author Email address: milos.cernak@idiap.ch (Milos Cernak) of the dysarthria level according to the Frenchay assessment score related to the larynx, and significant improvement is obtained for reading speech task. The proposed characterisation of VQ might also be applied to other kinds of pathological speech.

We investigate a vocoder based on artificial neural networks using a phonological speech representation. Speech decomposition is based on the phonological encoders, realised as neural network classifiers, that are trained for a particular language. The speech reconstruction process involves using a Deep Neural Network (DNN) to map phonological features posteriors to speech parameters -line spectra and glottal signal parameters -followed by LPC resynthesis. This DNN is trained on a target voice without transcriptions, in a semisupervised manner. Both encoder and decoder are based on neural networks and thus the vocoding is achieved using a simple fast forward pass. An experiment with French vocoding and a target male voice trained on 21 hour long audio book is presented. An application of the phonological vocoder to low bit rate speech coding is shown, where transmitted phonological posteriors are pruned and quantized. The vocoder with scalar quantization operates at 1 kbps, with potential for lower bit-rate.Index Terms-Parametric vocoding, low bit rate speech coding, phonology.

Correct pronunciation is known to be the most difficult part to acquire for (native or non-native) language learners. The accented speech is thus more variable, and standard Automatic Speech Recognition (ASR) training approaches that rely on intermediate phone alignment might introduce errors during the ASR training. With end-to-end training we could alleviate this problem. In this work, we explore the use of multi-task training and accent embedding in the context of end-to-end ASR trained with the connectionist temporal classification loss. Comparing to the baseline developed using conventional ASR framework exploiting time-delay neural networks trained on accented English, we show significant relative improvement of about 25% in word error rate. Additional evaluation on unseen accent data yields relative improvements of of 31% and 2% for New Zealand English and Indian English, respectively.

Abstract-Recent work in text to speech synthesis has pointed to the benefit of using a continuous pitch estimate; that is, one that records pitch even when voicing is not present. Such an approach typically requires interpolation. The purpose of this paper is to show that a continuous pitch estimation is available from a combination of otherwise well known techniques. Further, in the case of an autocorrelation based estimate, the continuous requirement negates the need for other heuristics to correct for common errors. An algorithm is suggested, illustrated, and demonstrated using a parametric vocoder.