No abstract

In order to produce an output speech signal that is more easily understood by a person with certain types of hearing loss, this system modifies the time base of particular portions of the signal. Voiced, voiceless, and silent portions of the input are identified based on average power, zero crossings, ORIGINAL SPEECH PROCESSED SPEECH •di• CENTER 5,305,332 43.72.Gy SPEECH DECODER FOR HIGH QUALITY REPRODUCED SPEECH THROUGH INTERPOLATION Kazunori Ozawa, assignor to NEC Corporation 2015