Example of PEMO processing of speech

An example for speech processing performed by the auditory model. On top, the waveforms of one utterance of the German word ``wiederholen'' from a female speaker in clean speech and in speech simulating noise at 10 dB SNR are shown. In the middle, the corresponding auditory-based representations after precessing are visualized. Large amplitudes are indicated by bright spots. At the bottom, the output of one single frequency channel (CF=780Hz) is shown. Peaks and ``undershoot'' are well maintained, whereas distortions due to noise are compressed.

Go back