Three-dimensional modeling of tongue during speech using MRI data

The tongue is the most important and dynamic articulator for speech formation, because of its anatomic aspects (particularly, the large volume of this muscular organ comparatively to the surrounding organs of the vocal tract) and also due to the wide range of movements and flexibility that are involved. In speech communication research, a variety of techniques have been used for measuring the three-dimensional vocal tract shapes. More recently, magnetic resonance imaging (MRI) becomes common; mainly, because this technique allows the collection of a set of static and dynamic images that can represent the entire vocal tract along any orientation. Over the years, different anatomical organs of the vocal tract have been modelled; namely, 2D and 3D tongue models, using parametric or statistical modelling procedures. Our aims are to present and describe some 3D reconstructed models from MRI data, for one subject uttering sustained articulations of some typical Portuguese sounds. Thus, we present a 3D database of the tongue obtained by stack combinations with the subject articulating Portuguese vowels. This 3D knowledge of the speech organs could be very important; especially, for clinical purposes (for example, for the assessment of articulatory impairments followed by tongue surgery in speech rehabilitation), and also for a better understanding of acoustic theory in speech formation.

Modelling and segmentation of the vocal tract during speech production by using deformable models in magnetic resonance images

The first and second authors would like to thank the support of the PhD grants with references SFRH/BD/28817/2006 and SFRH/PROTEC/49517/2009, respectively, from Fundação para a Ciência e Tecnol ogia (FCT). This work was partially done in the scope of the project “Methodologies to Analyze Organs from Complex Medical Images – Applications to Fema le Pelvic Cavity”, wi th reference PTDC/EEA- CRO/103320/2008, financially supported by FCT.

Analysis of tongue shape and motion in speech production using statistical modeling

The mechanisms of speech production are complex and have been raising attention from researchers of both medical and computer vision fields. In the speech production mechanism, the articulator’s study is a complex issue, since they have a high level of freedom along this process, namely the tongue, which instigates a problem in its control and observation. In this work it is automatically characterized the tongues shape during the articulation of the oral vowels of Portuguese European by using statistical modeling on MR-images. A point distribution model is built from a set of images collected during artificially sustained articulations of Portuguese European sounds, which can extract the main characteristics of the motion of the tongue. The model built in this work allows under standing more clearly the dynamic speech events involved during sustained articulations. The tongue shape model built can also be useful for speech rehabilitation purposes, specifically to recognize the compensatory movements of the articulators during speech production.

Phonological and articulation treatment approaches in Portuguese children with speech and language impairments: a randomized controlled intervention study

Background: In Portugal, the routine clinical practice of speech and language therapists (SLTs) in treating children with all types of speech sound disorder (SSD) continues to be articulation therapy (AT). There is limited use of phonological therapy (PT) or phonological awareness training in Portugal. Additionally, at an international level there is a focus on collecting information on and differentiating between the effectiveness of PT and AT for children with different types of phonologically based SSD, as well as on the role of phonological awareness in remediating SSD. It is important to collect more evidence for the most effective and efficient type of intervention approach for different SSDs and for these data to be collected from diverse linguistic and cultural perspectives. Aims: To evaluate the effectiveness of a PT and AT approach for treatment of 14 Portuguese children, aged 4.0–6.7 years, with a phonologically based SSD. Methods & Procedures: The children were randomly assigned to one of the two treatment approaches (seven children in each group). All children were treated by the same SLT, blind to the aims of the study, over three blocks of a total of 25 weekly sessions of intervention. Outcome measures of phonological ability (percentage of consonants correct (PCC), percentage occurrence of different phonological processes and phonetic inventory) were taken before and after intervention. A qualitative assessment of intervention effectiveness from the perspective of the parents of participants was included. Outcomes & Results: Both treatments were effective in improving the participants’ speech, with the children receiving PT showing a more significant improvement in PCC score than those receiving the AT. Children in the PT group also showed greater generalization to untreated words than those receiving AT. Parents reported both intervention approaches to be as effective in improving their children’s speech. Conclusions & Implications: The PT (combination of expressive phonological tasks, phonological awareness, listening and discrimination activities) proved to be an effective integrated method of improving phonological SSD in children. These findings provide some evidence for Portuguese SLTs to employ PT with children with phonologically based SSD

Differences in Mismatch Negativity (MMN) response to Pure-tone and Speech sounds in normal subjects: an additional explanation

The relation of automatic auditory discrimination, measured with MMN, with the type of stimuli has not been well established in the literature, despite its importance as an electrophysiological measure of central sound representation. In this study, MMN response was elicited by pure-tone and speech binaurally passive auditory oddball paradigm in a group of 8 normal young adult subjects at the same intensity level (75 dB SPL). The frequency difference in pure-tone oddball was 100 Hz (standard = 1 000 Hz; deviant = 1 100 Hz; same duration = 100 ms), in speech oddball (standard /ba/; deviant /pa/; same duration = 175 ms) the Portuguese phonemes are both plosive bi-labial in order to maintain a narrow frequency band. Differences were found across electrode location between speech and pure-tone stimuli. Larger MMN amplitude, duration and higher latency to speech were verified compared to pure-tone in Cz and Fz as well as significance differences in latency and amplitude between mastoids. Results suggest that speech may be processed differently than non-speech; also it may occur in a later stage due to overlapping processes since more neural resources are required to speech processing.

Coding mode decision algorithm for binary descriptor coding

In visual sensor networks, local feature descriptors can be computed at the sensing nodes, which work collaboratively on the data obtained to make an efficient visual analysis. In fact, with a minimal amount of computational effort, the detection and extraction of local features, such as binary descriptors, can provide a reliable and compact image representation. In this paper, it is proposed to extract and code binary descriptors to meet the energy and bandwidth constraints at each sensing node. The major contribution is a binary descriptor coding technique that exploits the correlation using two different coding modes: Intra, which exploits the correlation between the elements that compose a descriptor; and Inter, which exploits the correlation between descriptors of the same image. The experimental results show bitrate savings up to 35% without any impact in the performance efficiency of the image retrieval task. © 2014 EURASIP.

Enhancing Multimodal Silent Speech Interfaces With Feature Selection

In research on Silent Speech Interfaces (SSI), different sources of information (modalities) have been combined, aiming at obtaining better performance than the individual modalities. However, when combining these modalities, the dimensionality of the feature space rapidly increases, yielding the well-known "curse of dimensionality". As a consequence, in order to extract useful information from this data, one has to resort to feature selection (FS) techniques to lower the dimensionality of the learning space. In this paper, we assess the impact of FS techniques for silent speech data, in a dataset with 4 non-invasive and promising modalities, namely: video, depth, ultrasonic Doppler sensing, and surface electromyography. We consider two supervised (mutual information and Fisher's ratio) and two unsupervised (meanmedian and arithmetic mean geometric mean) FS filters. The evaluation was made by assessing the classification accuracy (word recognition error) of three well-known classifiers (knearest neighbors, support vector machines, and dynamic time warping). The key results of this study show that both unsupervised and supervised FS techniques improve on the classification accuracy on both individual and combined modalities. For instance, on the video component, we attain relative performance gains of 36.2% in error rates. FS is also useful as pre-processing for feature fusion. Copyright © 2014 ISCA.

H.264/AVC backward compatible Bit-Depth scalable video coding

As high dynamic range video is gaining popularity, video coding solutions able to efficiently provide both low and high dynamic range video, notably with a single bitstream, are increasingly important. While simulcasting can provide both dynamic range videos at the cost of some compression efficiency penalty, bit-depth scalable video coding can provide a better trade-off between compression efficiency, adaptation flexibility and computational complexity. Considering the widespread use of H.264/AVC video, this paper proposes a H.264/AVC backward compatible bit-depth scalable video coding solution offering a low dynamic range base layer and two high dynamic range enhancement layers with different qualities, at low complexity. Experimental results show that the proposed solution has an acceptable rate-distortion performance penalty regarding the HDR H.264/AVC single-layer coding solution.

Perceptually driven video error protection using a distributed source coding approach

In video communication systems, the video signals are typically compressed and sent to the decoder through an error-prone transmission channel that may corrupt the compressed signal, causing the degradation of the final decoded video quality. In this context, it is possible to enhance the error resilience of typical predictive video coding schemes using as inspiration principles and tools from an alternative video coding approach, the so-called Distributed Video Coding (DVC), based on the Distributed Source Coding (DSC) theory. Further improvements in the decoded video quality after error-prone transmission may also be obtained by considering the perceptual relevance of the video content, as distortions occurring in different regions of a picture have a different impact on the user's final experience. In this context, this paper proposes a Perceptually Driven Error Protection (PDEP) video coding solution that enhances the error resilience of a state-of-the-art H.264/AVC predictive video codec using DSC principles and perceptual considerations. To increase the H.264/AVC error resilience performance, the main technical novelties brought by the proposed video coding solution are: (i) design of an improved compressed domain perceptual classification mechanism; (ii) design of an improved transcoding tool for the DSC-based protection mechanism; and (iii) integration of a perceptual classification mechanism in an H.264/AVC compliant codec with a DSC-based error protection mechanism. The performance results obtained show that the proposed PDEP video codec provides a better performing alternative to traditional error protection video coding schemes, notably Forward Error Correction (FEC)-based schemes. (C) 2013 Elsevier B.V. All rights reserved.

Characterisation of noisy speech channels in 2G and 3G mobile networks

As the wireless cellular market reaches competitive levels never seen before, network operators need to focus on maintaining Quality of Service (QoS) a main priority if they wish to attract new subscribers while keeping existing customers satisfied. Speech Quality as perceived by the end user is one major example of a characteristic in constant need of maintenance and improvement. It is in this topic that this Master Thesis project fits in. Making use of an intrusive method of speech quality evaluation, as a means to further study and characterize the performance of speech codecs in second-generation (2G) and third-generation (3G) technologies. Trying to find further correlation between codecs with similar bit rates, along with the exploration of certain transmission parameters which may aid in the assessment of speech quality. Due to some limitations concerning the audio analyzer equipment that was to be employed, a different system for recording the test samples was sought out. Although the new designed system is not standard, after extensive testing and optimization of the system's parameters, final results were found reliable and satisfactory. Tests include a set of high and low bit rate codecs for both 2G and 3G, where values were compared and analysed, leading to the outcome that 3G speech codecs perform better, under the approximately same conditions, when compared with 2G. Reinforcing the idea that 3G is, with no doubt, the best choice if the costumer looks for the best possible listening speech quality. Regarding the transmission parameters chosen for the experiment, the Receiver Quality (RxQual) and Received Energy per Chip to the Power Density Ratio (Ec/N0), these were subject to speech quality correlation tests. Final results of RxQual were compared to those of prior studies from different researchers and, are considered to be of important relevance. Leading to the confirmation of RxQual as a reliable indicator of speech quality. As for Ec/N0, it is not possible to state it as a speech quality indicator however, it shows clear thresholds for which the MOS values decrease significantly. The studied transmission parameters show that they can be used not only for network management purposes but, at the same time, give an expected idea to the communications engineer (or technician) of the end-to-end speech quality consequences. With the conclusion of the work new ideas for future studies come to mind. Considering that the fourth-generation (4G) cellular technologies are now beginning to take an important place in the global market, as the first all-IP network structure, it seems of great relevance that 4G speech quality should be subject of evaluation. Comparing it to 3G, not only in narrowband but also adding wideband scenarios with the most recent standard objective method of speech quality assessment, POLQA. Also, new data found on Ec/N0 tests, justifies further research studies with the intention of validating the assumptions made in this work.

Replication vs erasure coding in data centric storage for wireless sensor networks

In-network storage of data in wireless sensor networks contributes to reduce the communications inside the network and to favor data aggregation. In this paper, we consider the use of n out of m codes and data dispersal in combination to in-network storage. In particular, we provide an abstract model of in-network storage to show how n out of m codes can be used, and we discuss how this can be achieved in five cases of study. We also define a model aimed at evaluating the probability of correct data encoding and decoding, we exploit this model and simulations to show how, in the cases of study, the parameters of the n out of m codes and the network should be configured in order to achieve correct data coding and decoding with high probability.

Performance Model for MRC Receivers with Adaptive Modulation and Coding in Rayleigh Fading Correlated Channels with Imperfect CSIT

RTUWO Advances in Wireless and Optical Communications 2015 (RTUWO 2015). 5-6 Nov Riga, Latvia.

Adaptive modeling and high quality spectral estimation for speech enhancement

In this work an adaptive modeling and spectral estimation scheme based on a dual Discrete Kalman Filtering (DKF) is proposed for speech enhancement. Both speech and noise signals are modeled by an autoregressive structure which provides an underlying time frame dependency and improves time-frequency resolution. The model parameters are arranged to obtain a combined state-space model and are also used to calculate instantaneous power spectral density estimates. The speech enhancement is performed by a dual discrete Kalman filter that simultaneously gives estimates for the models and the signals. This approach is particularly useful as a pre-processing module for parametric based speech recognition systems that rely on spectral time dependent models. The system performance has been evaluated by a set of human listeners and by spectral distances. In both cases the use of this pre-processing module has led to improved results.

Speech as the basic interface for assistive technology

Speech interfaces for Assistive Technologies are not common and are usually replaced by others. The market they are targeting is not considered attractive and speech technologies are still not well spread. Industry still thinks they present some performance risks, especially Speech Recognition systems. As speech is the most elemental and natural way for communication, it has strong potential for enhancing inclusion and quality of life for broader groups of users with special needs, such as people with cerebral palsy and elderly staying at their homes. This work is a position paper in which the authors argue for the need to make speech become the basic interface in assistive technologies. Among the main arguments, we can state: speech is the easiest way to interact with machines; there is a growing market for embedded speech in assistive technologies, since the number of disabled and elderly people is expanding; speech technology is already mature to be used but needs adaptation to people with special needs; there is still a lot of R&D to be done in this area, especially when thinking about the Portuguese market. The main challenges are presented and future directions are proposed.

Automatic syllabification for danish text-to-speech systems

In this paper, a rule-based automatic syllabifier for Danish is described using the Maximal Onset Principle. Prior success rates of rule-based methods applied to Portuguese and Catalan syllabification modules were on the basis of this work. The system was implemented and tested using a very small set of rules. The results gave rise to 96.9% and 98.7% of word accuracy rate, contrary to our initial expectations, being Danish a language with a complex syllabic structure and thus difficult to be rule-driven. Comparison with data-driven syllabification system using artificial neural networks showed a higher accuracy rate of the former system.