Widex study links natural-sounding AI denoising to better speech understanding

A study of Widex’s Clarity Boost AI noise-reduction technology found that preserving natural sound quality improved speech understanding in challenging noisy conditions. Hearing Review also reports that listeners experienced less effort and greater comfort, including in situations where understanding was already strong.

In the difficult listening conditions described in the article, speech intelligibility was about 50%. Clarity Boost improved speech understanding in those conditions, while participants also reported reduced listening effort and better hearing comfort. The article says the study found lower neural markers of listening effort as well, including when intelligibility was high.

The results were different in conditions where speech understanding was already very good. There, intelligibility was at least 90%, and the two programs did not produce a difference in speech understanding. Participants nevertheless said listening felt easier and more comfortable with Clarity Boost.

The article presents this as a comfort and effort finding distinct from an improvement in word understanding. Hearing Review describes the study’s significance in terms broader than word recognition alone. It says hearing-aid success also involves natural sound, comfort while listening and the mental effort required to follow conversations.

In that account, understanding what someone says is one part of the experience, rather than the only measure of it. The article also notes a possible difficulty with hearing-aid processing: systems may change sound or introduce artifacts. Those changes can increase listening effort and fatigue, according to the article.

The study’s reported results therefore address both speech understanding in noisy conditions and the perceived effort and comfort associated with listening. For the high-intelligibility conditions, Hearing Review says listening effort was lower even though speech understanding did not differ between programs.

It describes these conditions as more realistic for real-world communication and as reflecting many everyday situations encountered by hearing-aid users. The article cites Letowski and Scharine (2017) and Smeds and colleagues (2015) in support of that context.

Taken together, the account reports an improvement in understanding under difficult listening conditions and a separate comfort advantage when intelligibility was already high.

It does not provide further study details in the article excerpt, but emphasizes that natural sound, comfort and listening effort are relevant alongside speech recognition when assessing hearing-aid performance.