VoiceChoice is a patented technology for self-fitting of hearing aids by Two Pi. Unlike most of the diagnostic and fitting technologies it was not designed for the audiologist – VoiceChoice was from the start designed and developed for the patient.
OTC Hearing aids – a price/benefit zero-sum game?
The advent of OTC (sold Over-The-Counter) self-fitting hearing aid category in the US marks a paradigm change. The decision to enable sale of hearing aids aside of the traditional channel has provoked a fierce discussions about safety and efficacy, cost of healthcare and the future of audiologist-profession.
However, the discussion on OTC hearing aids assumed they wil not be subject to innovation and that the devices will be same as those sold in package with expert-driven fitting – less the diagnostics and the fitting service. The caution is understandable!
VoiceChoice proves that OTC devices don’t necessarily have to give up benefits – in order to make hearing aids more accessible.
VoiceChoice – Everyone is an Expert, Concerning Own Preferences
Sophisticated formulas and deep expertise are involved in the fitting of hearing aids, but there is one overriding criterium for the quality of fitting – the patient has to like the outcome, feel a psychological confirmation of the prescription. This is the reason why the expert-driven fitting is traditionally closed with assessing patient’s subjective preference by paired comparisons of slightly differing prescriptions.
The method of paired comparisons celebrates a revival today – now connected with smartphones running Machine Learning algorithms, it automatically tunes the hearing aid parameters to match the preference of the user. Widex, a Danish hearing aid manufacturer, introduced first such hearing aid, fine-tuned by A/B decision of the user. During the use the patient can provide her liking or disliking for a particular “sound” of the device – and the smartphone app statistically calculates the hearing-aid settings to fit the user’s preference.
VoiceChoice employs a similar idea but with two crucial differences:
- VoiceChoice doesn’t require an initial fitting based on diagnostics and involvement of an expert. (Its indication is limited to mild-to-moderate hearing loss, reducing so the range of available prescriptions.)
- VoiceChoice uses an audio streaming from smartphone to hearing aids to provide A/B samples.
Technical background of VoiceChoice:
VoiceChoice iteratively presents pairs of speech streams, each of them pre-processed with an assigned Wide Dynamic Range Compressor (WDRC) using different parameters. These speech streams – visually represented by two animated “speakers” – correspond to different signal processing parameters, A and B.
The sample audio is a calibrated recording of speech, from a speaker standing in front of the “listener”. Correction for hearing aid’s acoustic properties is included into processing on the smartphone.
The processed sample speech is provided in-situ using actual hearing aids, avoiding any deviation in sensitivity of smartphone speakers or audiometric headphones. What-You-Hear-Is-What-You-Get!
The VoiceChoice learning algorithm searches in multidimensional parameter space for the sweet-spot. This optimal amplification setting is inferred from the most desirable speech perception of the particular user.
The user is not confronted with the demanding before/after decision, he makes decisions in the intuitive speech realm – deciding between the visually presented left or right speaker. He also doesn’t have to rate the quality. (The answer: “I like the sound A 70% better than the sound B!” is not an intuitive statement and demands a significant cognitive effort.)
At the end of the procedure, the optimal amplification parameters, calculated in the VoiceChoice smartphone app are updated in the hearing aids – they are immediately ready for use.
…and even more advantages:
The VoiceChoice procedure can be repeated by the user at any time during the acclimatization and rehabilitation process – ensuring optimal hearing experience over long periods of use.
Because the VoiceChoice self-fitting routine is simple and non-stigmatizing, it gives users complete ownership of the prescriptive procedure.