> There is nothing in the ear to detect relative phase between different
> frequencies. There is some capability of detecting relative phase
> between the same frequencies arriving at both ears. This explanation
> doesn't apply binaural hearing.
Right. Under dynamic conditions (voice, music, etc.), the ear is relatively
insensitive to phase. The phase at 5kHz with respect
to 1 kHz goes completely unnoticed in ABX double-blind comparisons. However,
under static, steady-tone conditions, this is not
always the case.
As an example of this, nearly every FM broadcaster (including classical music
stations) employ an all-pass phase scrambling network
at the input to the audio processing stages. The effect is to severely
phase-rotate the audio to the extent that positive and
negative modulation peaks are equal in amplitude. Symmetrical processing of
the wave-form allows higher modulation (loudness)
densities before the artifacts become apparent. To a lesser degree, AM
broadcast stations utilize this technique, but the preferred
method is the use of a peak detector to gently flip or rotate the audio phase
depending on the audio passage. The result is that
maximum positive peak modulation at more than 125% is possible and maximum
loudness is maintained. Since music is nearly
symmetrical anyway, the purpose of both the phase scrambler and phase rotator
circuits applies primarily to vocal passages where the
source energy is oftentimes extremely asymmetrical, particularly on male voices.
I have digressed more than necessary, but the point remains: in dynamic audio
applications, the ear cannot distinguish between a
phase linear source and severely phase distorted source.
-Paul, W9AC
|