CLE PHONICS: Intelligent Electronic Device for Measuring Speech Comprehension

Main issues and objectives

PHONICS aims to extend the application scope of a tool developed in an AGILE-IT project, which involved the same partners. This first project made it possible to develop an automatic presetting solution for hearing aids in order to optimize and guarantee their use by patients suffering from presbycusis. This system will reduce the duration of the adaptation phase to hearing aids by evaluating the confusions in speech intelligibility and comprehension due to presbycusis.

Application of the comprehension measurement tool to American English

The first objective will be to extend the use of the AGILE-IT tool to American English. To do this, we will first replicate to the intelligibility and comprehension tests developed initially for French with English-speaking listeners by creating linguistic material adapted to the characteristics of English.

Applying the Competency Measurement Tool to Professional Deafness

The physiological characteristics of Professional Deafness are different from those of presbycusis. In particular, there are cochlear dead zones, that is to say the deterioration of certain parts of the Ciliated cells prevents tonotopyc coding. These dead zones are generally accompanied by tinnitus as well as a phenomenon of hyperacusis creating problems to understand speech in silence and in noise. Concerning these symptoms, they result in an audiogram showing a neuro-sensorial loss in the high frequencies and a localized notch in the 4 kHz region. The audiogram profile depends on the trauma and evolves over time, in particular the frequencies neighboring the region most affected can be affected gradually. To tackle this new problem, the protocol defined for the measurement of understanding in the case of presbycusis will be replicated by integrating the characteristics of the deafnesses by traumatism in professional environments. Consequently, the simulation of these types of deafness will be adapted according to the new acoustic parameters involved. The intelligibility and comprehension tests will be proposed with this new simulation in order to obtain scores corresponding to this deafness which will then be subjected on ASR, a speech recognition system.

Application of the Comprehension Measurement Tool to the Evaluation of Rehabilitation of Pathological Voices

The two objectives presented above were concerned with the comprehension disorders related to hearing pathologies, and the difficulties of understanding can also be a consequence of production difficulties related to pathologies of the phonatory apparatus.

Improving the intelligibility of pathological speech (cases of dysarthria, cases of ablation of the larynx following a cancer, etc.) is an important challenge in the field of health since these disorders induce a severe handicap of communication in everyday life. For example, patients with dysarthria (central articular speech disorder) have the following symptoms: joint difficulties, impaired speech rate, jerky rhythm, frequent alteration of the voice timbre, etc.

In order to manage dysarthria and other disorders of speech production, it is possible to reduce the degree of deviation from the speech production model on prosodic and phonetic parameters. In practice, this idea is difficult in view of the large number of targets to be corrected and the variability of speech production and perception methods. We are also confronted to the limitations of what the person can recover on the sensori-motor aspects. Throughout rehabilitation, practitioners are led to judge the consequences of treatment on the quality of speech production. These judgments are difficult to achieve because they are biased due to the doctor's habituation to pronunciation defects.

In this context, an objective and reliable measure of intelligibility and understanding would be very useful for improving the patient follow-ups. The results obtained could then be integrated into clinical reports, which would make it possible to significantly improve rehabilitation.

Collaborations

People involved in SAMOVA team

Support

Scheduled

  • Start time : 15th june 2015
  • End time : 31th december 2018