It may be that audio group delays must be so large to be perceived that it is
extraordinary to subjectively experience them, especially on their own.
Ordinarily, they will be associated with severe amplitude-frequency distortion.
I came across this reference at http://www.trueaudio.com/post_010.htm:
One of the most widely referenced reports of the audibility of group delay is:
Blauert, J. and Laws, P "Group Delay Distortions in Electroacoustical
Systems"
Journal of the Acoustical Society of America
Volume 63, Number 5, pp. 1478-1483 (May 1978)
Blauert and Laws report approximately the following thresholds for audibility:
Frequency Threshold of Audibility
8 kHz 2 msec
4 kHz 1.5 msec
2 kHz 1 msec
1 kHz 2 msec
500 Hz 3.2 msec
It's interesting that the most sensitive area is the speech spectrum. But
1 or 2 cycles of delay is still quite large.
The point is that group delay distortion is audible, if it's large
enough.
73, Tom VE3MEO
----- Original Message -----
From: Andy Green
To: [email protected]
Sent: Saturday, February 03, 2007 6:11 AM
Subject: Re: [soft_radio] Re: Tecnical discussion
----- Original Message -----
From: "Tom Holden" <[EMAIL PROTECTED]>
Subject: Re: [soft_radio] Re: Tecnical discussion
>snip< could not hear a difference on my notebook's poor sound system that
>would allow me to pick out which was which but I don't think we would hear
>any anyway on a simple steady-state signal such as this. It's a poor
>example. Group delay distortion is more discernable on transients and more
>complex waveforms.
>
> Why do finer multi-driver speaker systems attempt to time-align the
> drivers? Why do telecommunications systems control group delay on speech
> channels? Why is group delay controlled in analog television systems, in
> digital transmission systems? Because it can be a discernible and
> objectionable distortion if it's large enough and the signal or
> information is susceptible to it. But group delay does not generate new
> components in the signal - it just shifts them around in time. New
> waveform but the component frequencies and amplitudes are unchanged.
>
> I agree that the assertion misstates the nature of group delay distortion.
>
> Tom VE3MEO
In broadcast audio processing group delay is deliberately changed - the
audiio is run through allpass filters that are typically third order. The
reason for this is that it tends to make an assymetrical audio signal more
symmetrical. (male voice is often very assymetrical).
When audio processing was a relatively new field in the 1970's and AM
broadcasting was predominant, audio processing designers would use a soft
switch to detect the polarity of the assymetry and flip the phase during a
quite passage. This was refined by Bob Orban in the legendary Optimod 9000
so that the phase was slowly flipped over a perios of a few seconds (if you
are interested how and why in more detail, the manual, with full circuit
description is available for free download at ftp://ftp.orban.com/9000A1/
Since the limit for AM overmodulation is only critical in the negative
direction, (as the carrier gets pinched off), modulating over 100% in the
positive direction ensured a louder signal!
It was discovered in later incarnations Orban 9100, 8100) that all this
fancy circuitry was not really worth the bother, and that simply scrambling
the phase of the incoming audio resulted in audio that was at least as loud.
However, once the audio had been peak limited and or clipped, the group
delay performance becomes critical. Any change in phase response from sub-1
HZ to 100's of kHz will result in random peak overshoots which are programme
dependant. If this occurs, then the modulation level would have to be
lowered to allow headroom for these overshoots. Thsi is the reason
Broadcasters (and I susupect Telco companies) try and keep group delays to a
minimum, and not because the general public can hear the effects of them. I
can barely detect the difference between the 'phase scrambler' in an audio
processor being switched in or out, and then, only with an A-B comparison.
With Hi FI speakers, the effect of minimising group delay is to primarily to
aid the creation of a strong stereo image, which is less critical of the
listener position, rather than the effect any absolute change to the 'sound'
of a piece of audio due to group delay.
Andy EI3HG
[Non-text portions of this message have been removed]
