Over the past decades, a tremendous amount of research has been done on the use of machine learning for speech processing applications, especially speech recognition. However, in the past few years, research has focused on utilizing deep learning for speech-related applications. This new area of machine learning has yielded far better results when compared to others in a variety of applications including speech, and thus became a very attractive area of research. This paper provides a thorough examination of the different studies that have been conducted since 2006, when deep learning first arose as a new area of machine learning, for speech applications. A thorough statistical analysis is provided in this review which was conducted by extracting specific information from 174 papers published between the years 2006 and 2018. The results provided in this paper shed light on the trends of research in this area as well as bring focus to new research topics.

/pdf/speech-recognition-using-deep-neural-networks-a-systematic-kp43ilkz5h.pdf

Speech Recognition Using Deep Neural Networks: A Systematic Review

Clinical examination of voice

Motor Speech Disorders: Advances in Assessment and Treatment

Similar cerebral networks in language, music and song perception.

Hypokinetic dysarthria (HD) occurs in 90% of Parkinson’s disease (PD) patients. It manifests specifically in the areas of articulation, phonation, prosody, speech fluency, and faciokinesis. We aimed to systematically review papers on HD in PD with a special focus on (1) early PD diagnosis and monitoring of the disease progression using acoustic voice and speech analysis, and (2) functional imaging studies exploring neural correlates of HD in PD, and (3) clinical studies using acoustic analysis to evaluate effects of dopaminergic medication and brain stimulation. A systematic literature search of articles written in English before March 2016 was conducted in the Web of Science, PubMed, SpringerLink, and IEEE Xplore databases using and combining specific relevant keywords. Articles were categorized into three groups: (1) articles focused on neural correlates of HD in PD using functional imaging (n = 13); (2) articles dealing with the acoustic analysis of HD in PD (n = 52); and (3) articles concerning specifically dopaminergic and brain stimulation-related effects as assessed by acoustic analysis (n = 31); the groups were then reviewed. We identified 14 combinations of speech tasks and acoustic features that can be recommended for use in describing the main features of HD in PD. While only a few acoustic parameters correlate with limb motor symptoms and can be partially relieved by dopaminergic medication, HD in PD seems to be mainly related to non-dopaminergic deficits and associated particularly with non-motor symptoms. Future studies should combine non-invasive brain stimulation with voice behavior approaches to achieve the best treatment effects by enhancing auditory-motor integration.

Speech disorders in Parkinson's disease: early diagnostics and effects of medication and brain stimulation.

This paper investigates the adaptation of Automatic Speaker Recognition (ASR) techniques to the pathological voice assessment (dysphonic voices). The aim of this study is to provide a novel method, suitable for keeping track of the evolution of the patient's pathology: easy-to-use, fast, non-invasive for the patient, and affordable for the clinicians. This method will be complementary to the existing ones - the perceptual judgment and the usual objective measurement (jitter, airflows...) which remain time and human resource consuming. The system designed for this particular task relies on the GMMbased approach, which is the state-of-the-art for speaker recognition. It is derived from the open source ASR tools (LIA_Spk- Det and ALIZE) of the LIA lab.Experiments conducted on a dysphonic corpus provide promising results, underlining the interest of such an approach and opening further research investigation.

/pdf/application-of-automatic-speaker-recognition-techniques-to-sybnq0tfbc.pdf

Application of Automatic Speaker Recognition techniques to pathological voice assessment (dysphonia)

Background: Postural correlates of vocal effort are rarely described in the literature, while they are extensively dealt with in speech therapy. Objectives: This study aims at determining whether body movement is a side effect of vocal effort or an integral part of communication effort behavior. The answer to this question is mainly based on correlations between posture and phonation. Method: Twenty healthy subjects participated in this study. They had to communicate with a listener under 3 conditions requiring different levels of vocal effort. Results: The vocal parameters increased and confirmed that the subjects had made a vocal effort. The kinematic parameters (amplitude and duration of body movement) increased with vocal effort. Lastly, vocal and kinematic characteristics were significantly correlated. Conclusion: The close correlation of posture with vocal production shows that movement is not a mere consequence of vocal effort. Posture and voice are coordinated in communication behavior, and each body segment plays its specific role in the vocal effort behavior.

/pdf/coordination-between-posture-and-phonation-in-vocal-effort-17190i76ho.pdf

Coordination between posture and phonation in vocal effort behavior.

Since perception tests are highly time-consuming, there is a need to automate as many operations as possible, such as stimulus generation, procedure control, perception testing, and data analysis. The computer-driven system we are presenting here meets these objectives. To achieve large flexibility, the tests are controlled by scripts. The system's core software resembles that of a lexical-syntactic analyzer, which reads and interprets script files sent to it. The execution sequence (trial) is modified in accordance with the commands and data received. This type of operation provides a great deal of flexibility and supports a wide variety of tests such as auditory-lexical decision making, phoneme monitoring, gating, phonetic categorization, word identification, voice quality, etc. To achieve good performance, we were careful about timing accuracy, which is the greatest problem in computerized perception tests.

PERCEVAL: a Computer-Driven System for Experimentation on Auditory and Visual Perception

https://hal.archives-ouvertes.fr/hal-01317198/document

How to manage sound, physiological and clinical data of 2500 dysphonic and dysarthric speakers?

To measure oral and nasal airflow during speech production, several conditions must be gathered. In this aim, we designed and built a pneumotachograph with particular care to optimise its response time, linearity and acoustical response. This flow meter is based on the grid flow meter principle with a small dead volume and specific linearization for the inhaled and exhaled airflow. A soft silicone rubber mask, pressed against the speaker’s face prevents air leakage, without hindering articulatory movements. The acoustical distortions of the speech sound through the device are remedied by an adapted signal processing from its transfer function.

https://hal.archives-ouvertes.fr/hal-00142982/document

Alain Ghio

Papers

Application of Automatic Speaker Recognition techniques to pathological voice assessment (dysphonia)

Coordination between posture and phonation in vocal effort behavior.

PERCEVAL: a Computer-Driven System for Experimentation on Auditory and Visual Perception

How to manage sound, physiological and clinical data of 2500 dysphonic and dysarthric speakers?

Evaluation of the acoustic and aerodynamic constraints of a pneumotachograph for speech and voice studies