Özgül Salor

Turkish speech corpora and recognition tools developed by porting SONIC: Towards multilingual speech recognition

This paper presents work on developing speech corpora and recognition tools for Turkish by porting SONIC, a speech recognition tool developed initially for English at the Center for Spoken Language Research of the University of Colorado at Boulder. The work presented in this paper had two objectives: The first one is to collect a standard phonetically-balanced Turkish microphone speech corpus for general research use. A 193-speaker triphone-balanced audio corpus and a pronunciation lexicon for Turkish have been developed. The corpus has been accepted for distribution by the Linguistic Data Consortium (LDC) of the University of Pennsylvania in October 2005, and it will serve as a standard corpus for Turkish speech researchers. The second objective was to develop speech recognition tools (a phonetic aligner and a phone recognizer) for Turkish, which provided a starting point for obtaining a multilingual speech recognizer by porting SONIC to Turkish. This part of the work was the first port of this particular recognizer to a language other than English; subsequently, SONIC has been ported to over 15 languages. Using the phonetic aligner developed, the audio corpus has been provided with word, phone and HMM-state level alignments. For the phonetic aligner, it is shown that 92.6% of the automatically labeled phone boundaries are placed within 20ms of manually labeled locations for the Turkish audio corpus. Finally, a phone recognition error rate of 29.2% is demonstrated for the phone recognizer.

Dynamic programming approach to voice transformation

This paper presents a voice transformation algorithm which modifies the speech of a source speaker such that it is perceived as if spoken by a target speaker. A novel method which is based on dynamic programming approach is proposed. The designed system obtains speaker-specific codebooks of line spectral frequencies (LSFs) for both source and target speakers. Those codebooks are used to train a mapping histogram matrix, which is used for LSF transformation from one speaker to the other. The baseline system uses the maxima of the histogram matrix for LSF transformation. The shortcomings of this system, which are the limitations of the target LSF space and the spectral discontinuities due to independent mapping of subsequent frames, have been overcome by applying the dynamic programming approach. Dynamic programming approach tries to model the long-term behaviour of LSFs of the target speaker, while it is trying to preserve the relationship between the subsequent frames of the source LSFs, during transformation. Both objective and subjective evaluations have been conducted and it has been shown that dynamic programming approach improves the performance of the system in terms of both the speech quality and speaker similarity.

A dynamic programming approach to voice transformation

This paper presents a voice transformation algorithm which modifies the speech of a source speaker such that it is perceived as if spoken by a target speaker. A novel method which is based on dynamic programming approach is proposed. The designed system obtains speaker-specific codebooks of line spectral frequencies (LSFs) for both source and target speakers. Those codebooks are used to train a mapping histogram matrix, which is used for LSF transformation from one speaker to the other. The baseline system uses the maxima of the histogram matrix for LSF transformation. The shortcomings of this system, which are the limitations of the target LSF space and the spectral discontinuities due to independent mapping of subsequent frames, have been overcome by applying the dynamic programming approach. Dynamic programming approach tries to model the long-term behaviour of LSFs of the target speaker, while it is trying to preserve the relationship between the subsequent frames of the source LSFs, during transformation. Both objective and subjective evaluations have been conducted and it has been shown that dynamic programming approach improves the performance of the system in terms of both the speech quality and speaker similarity. 2006 Elsevier B.V. All rights reserved.

Speech conversion using MELP speech coding algorithm

In this work, the MELP (mixed excitation linear prediction) speech coding algorithm has been used for speech conversion. Speech conversion aims to modify the speech of one speaker such that the modified speech sounds as if spoken by another speaker. Speech modeling of MELP has been used to derive a mapping the between the speech models of the two speakers. We have obtained a mapping which provides a context-free speech conversion. We have mainly considered the spectral properties of the speakers. Using the 230 sentences of the two speakers, a mapping between the 4-stage vector quantization indexes for line spectral frequencies (LSF) of the two speakers have been obtained. Two different methods have been proposed to obtain a codebook for the second speaker from this mapping and both have been applied in addition to pitch modification during synthesis. The first method replaces the LSF index of the first speaker with that of the second speaker, which appears the most, during training. The second method uses the weighted average from the histogram of the second speaker that corresponds to the index of the first speaker, to form a new LSF codebook for the second speaker. Subjective ABX listening tests have been carried out and the correct speaker perception rate has been obtained as 70% and 65% for the first and the second spectral conversion methods respectively.