[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Granular synthesis and auditory segmentation


>No, I haven't tried this, because I understood from your
>earlier decriptions that the sounds you used (human voices)
>included frequencies below 3 kHz as well.

        ... Sorry, I did not understand your requirement !
        ... Thought your ...
        "unambiguously demonstrate[s] temporal processing in humans in
        the 3 to 5 kHz range"
                ... was meant as an upper limit, not a 3 - 5 kHz bandpass!

        ... Though all speech signals, with the exception of (resonant)
        singing, have frequencies below 2 kHz and are thus processed
        for ITD and localization features within the range of the HALFwave
        IHC properties discussed earlier.

        ... But, one typically doesn't need to acoustically locate opera singers
        as this would greatly insult the resplendent effort at stage sets and
        costumes ... and, the acoustic mixing effects of a great opera hall try
        purposely to allow resonant combinations to fill the environment.

>I did not study your ideas in detail, but I got the impression that it is
>kind of (interesting, though) precedence effect "on top of" [B]MLD
>(binaural masking level differences) and binaural phase detection.

        ... A BMLD test is always done using headphones.
        ... However, the posted experiment always uses loudspeakers
        in a room of arbitrary acoustic "liveliness" (reverb), i.e., this is a
        quite realistic test which approximates a true CPE environment.

        ... A BMLD test demonstrates how an (in-phase) binaural test signal
        masked by (in-phase) wideband noise can (once again) be perceived
        as audible by simply presenting the test signal in-phase at one ear
        and inverted-phase at the other ear.

        ... BMLD is not demonstrated by the posted experiment

>Binaural phase detection too has been reported to vanish
>around these transition frequencies. Hence, anything
>"on top of" MLD and binaural phase detection most likely
>vanishes too at that transition frequency.

        ... Actually, as the frequency of the (in-phase/inverted-phase)
        Binaural test signal is increased, TWO images of the test frequency
        procede from each ear towards the nose and are perceived as being
        "outside" the head.
        ... This effect is more easily perceived if you turn-off the mixed noise.

        ... This Binaural effect goes well beyond 2 kHz up to near 10 kHz
        depending on the hearing of the test subject.
        ... But, this Binaural test is not a fair test for just now as I am well
        aware of exactly why we perceive that dual image and that is a
        different topic ... but, it is also a great test that helps demonstrate
        our localization abilities used during CPE.

>Did you try using steeply high-pass filtered "voices"
>such that only 3 kHz and higher frequencies remained?
>Otherwise, your experiments simply do not apply to my
>questioning the neural "temporal processing" above
>3 kHz in terms of psychophysically observable effects.

        ... No, I did not.
        ... And, I have a reason for (probably) never doing so.
        ... I have a fundamentally different model through which I explore
        the connection between physiology, neural nets and the perceived
        effects we call "pyschoacoustic".
        ... I do not consider sine waves to be valid test signals.
        ... 99.9% of Nature provides complex, bursty acoustic signals.
        ... I simply realize that frequency analysis is a mathematical tool.
        ... And, that tool stays on the my shelf.
        ... I do not try to force a "round blocks into square holes".
        ... Does handling complex, bursty acoustic signals in the direct,
        time-locked manner in which they are produced excite anyone?

Best Regards,
Rich Fabbri

McGill is running a new version of LISTSERV (1.8d on Windows NT). 
Information is available on the WEB at http://www.mcgill.ca/cc/listserv