This study investigates the possible benefits of using binaural spatialization techniques in digital wireless microphone systems for hearing aids. Speech intelligibility tests, speaker localization tests and preference tests are performed. The results of a diotic (current rendering) and a binaural (suggested rendering) rendering are compared.
The signal processing laboratory (LTS2) of the Swiss Federal Institute of Technology (EPFL) has developed a new feature of sound rendering in hearing aids. This is a collaboration between EPFL and the Swiss company Phonak Communications AG. The new functionality works with the range of Roger products from Phonak, a new generation of FM systems that use a digital transmission. FM technology refers to a type of wireless system that helps people better understand speech in noisy situations. FM systems commonly work together with a user's hearing aids, although systems are also available for those with otherwise normal hearing (such as people who suffer from APD, ADHD etc.). An FM system works like this: the person speaking wears or holds a transmitter microphone, or the transmitter is placed in the middle of the group (picking up speech from all around). Using harmless radio waves, the FM system sends speech signal(s) to the listener, who wears a tiny FM receiver behind the ear. The new feature of such systems performs processing of the speaker's speech signal so that it renders information related to their spatial location. This so-called sound "spatialization" is a natural property of the human binaural auditory system, which allows us to localize sounds. However, this sound spatialization is not delivered by the current FM systems. Thanks to the new technology from EPFL and Phonak, it is now possible to include a binaural spatialization of the speech signal in the Roger product of Phonak hearing devices.
Study Type
INTERVENTIONAL
Allocation
NON_RANDOMIZED
Purpose
SUPPORTIVE_CARE
Masking
NONE
Enrollment
40
The only intervention consists in applying a specific processing on some recorded speech signals, and comparing the performance obtained with such processed samples with ones that have not been processed. The applied processing is a binaural spatialization method that consists in filtering an original audio signal to get a left and right versions (for the two ears). The binaural rendering gives the impression that the speech signal (and thus the speaker) is located in a desired position in the environment.
Ecole Polytechnique Fédérale de Lausanne
Lausanne, Switzerland
Speech Intelligibility
Speech recognition score (%) For each group: average of the SRS over all subjects in the group. The SRS correspond to the number of understood words in a sequence of sentences (French HINT database) mixed with several level of masking noise (speech-shaped noise). Some are played diotically, the other are spatialized in various directions. The goal is to ensure that the spatialization processing does not degrade the understanding of the speech.
Time frame: 1 day of the experiment
Speaker's Localization
Localization error (in number of spatial sectors) For each group: average localization error over all subjects in the group. It is the difference between the actual spatial sector and the one reported by the listeners. There were 5 spatial sectors, and 9 possible locations. For instance: if the stimuli is played in sector 4, and the listener perceives it in 2, then the localization error is \|4-2\| = 2. The goal is to compare the localization error in 3 conditions: 1/ with no hearing aids (reference of natural localization) and no spatialization 2/ with hearing aids and standard fittings and no spatialization, and 3/ with spatialization applied.
Time frame: 1 day of the experiment
Listener's Subjective Preference
Listeners compare two audiovisual stimuli (diotic and spatialized) and Indicate their preference between binaural diotic (no spatialization), spatialized stimuli, or no preference. Results are given as the percentage of participant for the 3 possible answers in each group.
Time frame: 1 day of the experiment
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.