Marc René Schädler(Carl von Ossietzky University Oldenburg)
This contributions describes the "IISPA" submission to the Hurricane Challenge 2.0. The challenge organizers called for submissions of speech signals processed with the aim to improve their intelligibility in adverse listening conditions.
They evaluated the submissions with matrix sentence tests in an international listening experiment. An intelligibility-improving signal processing approach (IISPA) inspired from research on speech perception of listeners with impaired hearing was designed. Its parameters were optimized with an objective intelligibility model, the simulation framework for auditory discrimination experiments (FADE). In FADE, a re-purposed automatic speech recognition (ASR) system is employed as a models for human speech recognition performance. The model predicted an improvement in speech recognition threshold (SRT) of approximately 5.0dB due to the optimized IISPA. The processed speech signals were evaluated in the Hurricane Challenge 2.0. The measured improvements were language-dependent: up to 4.8dB for the Spanish test, up to 3.8dB for the German test, and up to 2.1dB for the English test. The results show on the one hand the potential of using an ASR-based speech recognition model to optimize an intelligibility-improving signal processing scheme, and on the other hand the need for thorough listening experiments.