Emotion transplantation through adaptation in HMM-based speech synthesis

This paper proposes an emotion transplantation method capable of modifying a synthetic speech model through the use of CSMAPLR adaptation in order to incorporate emotional information learned from a different speaker model while maintaining the identity of the original speaker as much as possible. The proposed method relies on learning both emotional and speaker identity information by means of their adaptation function from an average voice model, and combining them into a single cascade transform capable of imbuing the desired emotion into the target speaker. This method is then applied to the task of transplanting four emotions (anger, happiness, sadness and surprise) into 3 male speakers and 3 female speakers and evaluated in a number of perceptual tests. The results of the evaluations show how the perceived naturalness for emotional text significantly favors the use of the proposed transplanted emotional speech synthesis when compared to traditional neutral speech synthesis, evidenced by a big increase in the perceived emotional strength of the synthesized utterances at a slight cost in speech quality. A final evaluation with a robotic laboratory assistant application shows how by using emotional speech we can significantly increase the students’ satisfaction with the dialog system, proving how the proposed emotion transplantation system provides benefits in real applications.

Improving Speech Interaction in Vehicles Using Context-Aware Information through A SCXML Framework

Speech Technologies can provide important benefits for the development of more usable and safe in-vehicle human-machine interactive systems (HMIs). However mainly due robustness issues, the use of spoken interaction can entail important distractions to the driver. In this challenging scenario, while speech technologies are evolving, further research is necessary to explore how they can be complemented with both other modalities (multimodality) and information from the increasing number of available sensors (context-awareness). The perceived quality of speech technologies can significantly be increased by implementing such policies, which simply try to make the best use of all the available resources; and the in vehicle scenario is an excellent test-bed for this kind of initiatives. In this contribution we propose an event-based HMI design framework which combines context modelling and multimodal interaction using a W3C XML language known as SCXML. SCXML provides a general process control mechanism that is being considered by W3C to improve both voice interaction (VoiceXML) and multimodal interaction (MMI). In our approach we try to anticipate and extend these initiatives presenting a flexible SCXML-based approach for the design of a wide range of multimodal context-aware HMI in-vehicle interfaces. The proposed framework for HMI design and specification has been implemented in an automotive OSGi service platform, and it is being used and tested in the Spanish research project MARTA for the development of several in-vehicle interactive applications.

Towards speaking style transplantation in speech synthesis

One of the biggest challenges in speech synthesis is the production of naturally sounding synthetic voices. This means that the resulting voice must be not only of high enough quality but also that it must be able to capture the natural expressiveness imbued in human speech. This paper focus on solving the expressiveness problem by proposing a set of different techniques that could be used for extrapolating the expressiveness of proven high quality speaking style models into neutral speakers in HMM-based synthesis. As an additional advantage, the proposed techniques are based on adaptation approaches, which means that they can be used with little training data (around 15 minutes of training data are used in each style for this paper). For the final implementation, a set of 4 speaking styles were considered: news broadcasts, live sports commentary, interviews and parliamentary speech. Finally, the implementation of the 5 techniques were tested through a perceptual evaluation that proves that the deviations between neutral and speaking style average models can be learned and used to imbue expressiveness into target neutral speakers as intended.

BioMet®Phon: A system to monitor phonation quality in the clinics

BioMet®Phon is a software application developed for the characterization of voice in voice quality evaluation. Initially it was conceived as plain research code to estimate the glottal source from voice and obtain the biomechanical parameters of the vocal folds from the spectral density of the estimate. This code grew to what is now the Glottex®Engine package (G®E). Further demands from users in laryngology and speech therapy fields instantiated the development of a specific Graphic User Interface (GUI’s) to encapsulate user interaction with the G®E. This gave place to BioMet®Phon, an application which extracts the glottal source from voice and offers a complete parameterization of this signal, including distortion, cepstral, spectral, biomechanical, time domain, contact and tremor parameters. The semantic capabilities of biomechanical parameters are discussed. Study cases from its application to the field of laryngology and speech therapy are given and discussed. Validation results in voice pathology detection are also presented. Applications to laryngology, speech therapy, and monitoring neurological deterioration in the elder are proposed.

A comparison of open-source segmentation architectures for dealing with imperfect data from the media in speech synthesis

Traditional Text-To-Speech (TTS) systems have been developed using especially-designed non-expressive scripted recordings. In order to develop a new generation of expressive TTS systems in the Simple4All project, real recordings from the media should be used for training new voices with a whole new range of speaking styles. However, for processing this more spontaneous material, the new systems must be able to deal with imperfect data (multi-speaker recordings, background and foreground music and noise), filtering out low-quality audio segments and creating mono-speaker clusters. In this paper we compare several architectures for combining speaker diarization and music and noise detection which improve the precision and overall quality of the segmentation.