Processing of Ear Signals by a
Cocktail-Party-Processor, Audio Samples

Deutsche Fassung

Used Cocktail-Party-Processor:

A detailed description can be found in the Dissertation (in German):

"Algorithms for direction specific Processing of Sound Signals -
the Realisation of a binaural Cocktail-Party-Processor-System",
Harald Slatky, Ruhr-University Bochum, 1992


Desired female speaker at the right (interaural time difference 400 µs)
Interfering male speaker in front (interaural time difference 0 µs)

Analysed Ear Signals:

The ear signals have been simulated for a "mean" human head. In order to determine a "mean" human head, the relationship between interaural time difference and interaural level difference has been measured at 3 persons and averaged over these 3 persons. (see Appendix C in Dissertation above)
The ear signals have been simulated under free field conditions without any wall reflections.

Signals of the Audio Samples:

Unprocessed signal: original ear signals of both speakers, summed up
Processed signal: output signal of the Cocktail-Party-Processor
Audio samples of the sound situation: Signals of desired speaker, interfering speaker, unprocessed signal, processed signal

Audio Samples

Level of the interfering speaker

Level of the desired speaker

Level difference desired/interfering speaker

Unprocessed signal

Processed signal

Audio samples of the sound situation

50 dB

50 dB

0 dB

50 dB

40 dB

-10 dB

50 dB

30 dB

-20 dB

50 dB

20 dB

-30 dB