Near-End Listening Enhancement in the Presence of Bandpass Noises

Abstract:
When using a mobile phone in acoustical background noise, the near-end listener perceives not only the (clean) far-end speech but also the ambient noise and thus experiences an increased listening effort and a possibly reduced speech intelligibility. Near-end listening enhancement processes the received far-end speech signal depending on the near-end background noise to improve intelligibility. However, in mobile phones it is often not possible to increase the audio power. In a previous contribution, the authors developed a recursive closed-form solution, which maximizes the Speech Intelligibility Index (SII) under the constraint of an unchanged average power of the audio signal. This solution, however, shows in bandpass noise environments a disadvantageous narrow bandpass characteristic of the processed speech even though the SII is optimal. Therefore, in this contribution, we analyze this algorithm and propose a new spectral weighting rule which prevents the narrow bandpass effect with only a marginal reduction in SII.