Listener Panel

There are various approaches to assessing the audio quality of music or other audio signals such as speech. For example, numerous perceptual models of audio quality have been developed, such as the Perceptual Evaluation of Audio Quality (PEAQ), the Hearing-Aid Speech Quality Index (HASQI), and similarly the Hearing-Aid Audio Quality Index (HAAQI). These models can use software to simulate the human ear, auditory models, hearing loss, and hearing aid processing. Commonly, a degraded audio signal is compared with a reference signal in the perceptual models, and correlations between key audio features are performed to produce an evaluation metric of audio quality (e.g., ranging from 0 to 1). The HAAQI will be used to score your systems, but it is also important to collect data regarding the subjective perceptual experiences of listeners with hearing loss. 

Selected systems will also be rated in terms of perceived audio quality by a panel of listeners with hearing loss; the ratings used by these participants are developed through sensory evaluation studies. This will ultimately mean that signal outputs will be judged directly by those listeners that your system has processed for, and this subjective data can further be used to improve and develop on existing perceptual models of audio quality, with these refined models to be used moving forward in further Cadenza challenges.

Further details of the listening tests to be used in the first Challenge can be found on the Listening tests page.