[linux-audio-user] Re: 192kHz

Wolfgang Woehl tito at rumford.de
Sun Jan 29 15:10:09 EST 2006


fons adriaensen <fons.adriaensen at skynet.be>:
> On Sun, Jan 29, 2006 at 01:40:46PM +0100, Wolfgang Woehl 
wrote:
> > I make 2 audio files with 400 and 500 hz sine tones and
> > play them along (out of ardour). There is clearly audible
> > beat frequency, a 100 hz hum and harmonics of that.

> Did you send the ardour output directly to jaaa or did the
> signal go via some analog path ? I just tried the same
> with two jaaa instances, one generating 400 Hz and the
> other 500, summed via jack and measured in either jaaa.
> No traces of 100 Hz.

ardour master bus -> spdif out / da -> pa / speakers
                  -> jaaa
So what I hear is what you say is called "intermodulation 
distortion" from the speakers. I don't see it in jaaa either. 
Point taken.

Now given the non-linearities others and you (below) describe 
and given this is about mixing discrete signals (not a 
summing recording of the audible range) and given there are 
instruments that emit frequencies or produce harmonics well 
above the audible range (violins, Brian Wilson 40 years ago) 
it makes sense to record it, doesn't it? With higher 
samplerates?

btw: I couldn't find the command line switch to give jaaa 
instances a jack-name in order to run more than 1, it doesn't 
seem to show up in jaaa -h?

> (Side remark: if your signals go through any analog
> hardware, it's better to avoid 100 Hz as a beat frequency.
> The mains frequency is 50 Hz in Europe, and I still have to
> see the first audio card that doesn't show up traces of 50,
> 100, 150, ... Hz in jaaa.)

Oh dear, yes.

> There are several different mechanisms at work here that
> can make you hear 100 Hz with your test.
>
> First, in order to hear anything at all, your signal has to
> be reproduced by a speaker, and these *do* produce
> intermodulation distortion. This can be quite noticeable in
> such a test, and in fact that's how it is measured. The
> sound card and amplifier will distort as well, but normally
> much less than any speaker.
>
> Second, you have chosen two frequencies that have a simple
> ratio (4/5). Human hearing will try to 'decode' them as
> a single signal, in this case as the harmonics of an 100
> Hz one. So you could *hear* 100 Hz even if it isn't present
> physically. The effect has been kwown for ages by organ
> builders, who use it to make 'fake' 32" stops.

I'm starting to "hear things" already,
http://hyperphysics.phy-astr.gsu.edu/hbase/sound/subton.html
notes a "trio for two flutes". That I would like to hear.

-- 
Wolfgang




More information about the Linux-audio-user mailing list