Speaking Style Conversion from Normal to Lombard Speech Using a Glottal Vocoder and Bayesian GMMs

Abstract

Speaking style conversion is the technology of converting natural speech signals from one style to another. In this study, we focus on normal-to-Lombard conversion. This can be used, for example, to enhance the intelligibility of speech in noisy environments. We propose a parametric approach that uses a vocoder to extract speech features. These features are mapped using Bayesian GMMs from utterances spoken in normal style to the corresponding features of Lombard speech. Finally, the mapped features are converted to a Lombard speech waveform with the vocoder. Two vocoders were compared in the proposed normal-to-Lombard conversion: a recently developed glottal vocoder that decomposes speech into glottal flow excitation and vocal tract, and the widely used STRAIGHT vocoder. The conversion quality was evaluated in two subjective listening tests measuring subjective similarity and naturalness. The similarity test results show that the system is able to convert normal speech into Lombard speech for the two vocoders. However, the subjective naturalness of the converted Lombard speech was clearly better using the glottal vocoder in comparison to STRAIGHT.

Cycle-consistent Adversarial Networks for Non-parallel Vocal Effort Based Speaking Style Conversion

“…In the baseline system with parallel data, a standard GMM is used as it was shown to compare well against DNNs and non-parametric Bayesian methods in an earlier study with the present data set [14].…”

“…Moreover, the amount of parallel training data, where the utterances in the source and target styles are from the same speaker speaking the same linguistic content, is limited. Our earlier work [14,15] also suggest that the limited availability of parallel data in normal and Lombard styles causes a bottleneck in system performance. This encourages the use of nonparallel mapping models within the parametric SSC framework.…”

“…Lombard speech corresponds to a speaking style that talkers naturally employ in noisy environments to improve intelligibility. We use a data driven parametric setup that uses a vocoder to extract speech features (see [14]). These features are mapped This study was funded by Academy of Finland grant nos.…”

“…Given this background, the overall goal of the current paper is to study the applicability of CycleGANs for the task of vocal effort based SSC (normal vs. Lombard), and to compare it to the standard INCA-based non-parallel approach and to our previous baseline system utilizing parallel data [14]. The systems are compared using subjective listening tests evaluating the success of style conversion and overall quality of the converted speech.…”

Augmented CycleGANs for Continuous Scale Normal-to-Lombard Speaking Style Conversion

“…SSC has been previously studied in whisper-to-normal conversion [3][4][5] and in normal-to-Lombard conversion [6][7][8]. In addition, a parametric approach to normal-to-Lombard SSC was recently explored in [9], where a vocoder was used to extract frame level features that were then transformed from normal to Lombard style using parallel data-driven mapping models, and then synthesized as speech in the target style using the same vocoder.…”