It is currently Thu Mar 28, 2024 4:07 am


Beyond Stereo to multiple Speakers

Speakers, amplifiers, headphones, multi-channel audio, reverb units, mixers, wiring, ...
  • Author
  • Message
Offline
User avatar

toplayer2

Member

  • Posts: 1071
  • Joined: Mon Jul 23, 2007 2:15 pm
  • Location: Michigan, USA

Re: Beyond Stereo to multiple Speakers

PostFri Oct 01, 2010 11:04 am

Martin,

Thank you for your detailed reply and analysis. I stipulate that my comments regarding multi-channel benefits are to be construed as applicable to dry sampled organs.

What I hear is the spurious generation of new frequencies. It is rather like two notes are turned into a nasty triad. These are what offend my ears and not the dissonance caused by constructive and destructive interfernce (beats) caused by tuning. Indeed, I personally enjoy highly dissonant harmonizations such as Penderecki's tone clusters as heard for example in "Threnody For The Victims Of Hiroshima", or thickly textured jazz chords. To my ears the IM heard in the demonstration is very noticeable as well as undesireable. I will repeat the demonstration using sine waves with the intervals tuned to Just Intonation. I will change the audio chain to a MOTU 24 io PCIe and Mackie HR824. Also I will include an example with the two notes routed to different audio channels and air mixed. If I am right, the differences should proove to be very audible and measureable. If I'm wrong, it won't be the first time nor the last.

Joe
Offline
User avatar

mdyde

Moderator

  • Posts: 15441
  • Joined: Fri Mar 14, 2003 1:19 pm
  • Location: UK

Re: Beyond Stereo to multiple Speakers

PostFri Oct 01, 2010 11:15 am

Hello Joe,

What I hear is the spurious generation of new frequencies. It is rather like two notes are turned into a nasty triad. These are what offend my ears and not the dissonance caused by constructive and destructive interfernce (beats) caused by tuning. Indeed, I personally enjoy highly dissonant harmonizations such as Penderecki's tone clusters as heard for example in "Threnody For The Victims Of Hiroshima", or thickly textured jazz chords. To my ears the IM heard in the demonstration is very noticeable as well as undesireable


Which specifically of the frequencies shown in my FFTs do you find noticeable/objectionable? One of the ones highlighted in red, or can you find some I missed? The red ones are certainly newly-introduced IMD frequencies, but I wouldn't have thought -45 dB would be especially noticeable or objectionable? Maybe you're particularly sensitive to those particular frequencies?

I will repeat the demonstration using sine waves with the intervals tuned to Just Intonation. I will change the audio chain to a MOTU 24 io PCIe and Mackie HR824. Also I will include an example with the two notes routed to different audio channels and air mixed.


Yes - that should certainly make it easier to see/measure the frequencies.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.
Offline
User avatar

bcollins

Member

  • Posts: 837
  • Joined: Sun Jun 04, 2006 10:09 pm
  • Location: USA, WA, Spokane

Re: Beyond Stereo to multiple Speakers

PostFri Oct 01, 2010 1:01 pm

I just want to mention that for some of us - we who use only dry samples, and use HW in a 'professional' installation, this is extremely important, and I appreciate Martin [and Joe] taking the time away from other responsibilities to do this research and analysis.
Bob Collins
Offline
User avatar

Jim Reid

Member

  • Posts: 1764
  • Joined: Mon May 17, 2004 6:50 pm
  • Location: Kauai, Hawaii

Re: Beyond Stereo to multiple Speakers

PostFri Oct 01, 2010 2:52 pm

IM Distortion Products:

Folks use Taylor Series expansion (see most calculus texts)
to determine IMD amplitudes and frequencies. The most simple approach
is to use two equal amplitude tones, sine waves, each within the same octave.

The third order IM products will be found at frequencies of 2fsub1-fsub2,
and 2fsub2-sf1 . Or, as Joe did, pick C at 525.25 Hz and the third above,
E at 659.26 (equal tempered tuning). The two generated 3rd order
IMs will be at 391.24 Hz; the other at 793.27 Hz. Notice that they
at an equal number of Hz above and below the frequencies of the
generating pair of tones. In this case 134.01 Hz. In fact, the
3rd IMs will be within the same octave of tones about the
generating fundamental pitch frequencies!

What about the IM products strength, or loudness compared
to the fundamental tones. They will be lower, unless the two generating
tones are very high in strength. At very loud amplitudes, the IM products can
approach the loudness of the fundamental tones; all as predicted by the
Taylor series expansion math.

More could be said, but all of this is clearly seen if having a look with a good
audio spectrum analyzer. Perhaps Joe can report such behavior after he performs
his described experiments.

View the sounds Joe has posted/will be with Windows Media Player using
the "stars and bars" display. This display shows a "poor man's"
spectrum display, low frequencies to the left, high to the right--
maybe when Joe uses pure tones (sine waves) the IM's will
be more clearly seen.
Jim Reid
Offline
User avatar

David Pinnegar

Member

  • Posts: 437
  • Joined: Thu Jan 03, 2008 7:13 pm
  • Location: Sussex UK

Re: Beyond Stereo to multiple Speakers

PostFri Oct 01, 2010 4:26 pm

bcollins wrote:I just want to mention that for some of us - we who use only dry samples, and use HW in a 'professional' installation, this is extremely important, and I appreciate Martin [and Joe] taking the time away from other responsibilities to do this research and analysis.

Hi!

This thread is beginning to be very fascinating. And thanks Martin for such excellently interesting graphic analysis. What is interesting is that that modulation components identified are only marginally above the noise floor, so it's perplexing.

What I suspect is that it's plain beat frequencies which are the problem and if I'm right it's probably only noticable at high volumes. Those of us who have tuned pianos and real pipes will perhaps be more familiar with the sounds than others, and we might just simply be expecting it.

My specific experience is of multi-channel dry reproduction with added reverberation and it's often the reverb unit that introduces the difference frequencies. Many of my sound sources are C-C# separated but some are single channel. I have compared an ear splitting Tuba in mono, really at a level which one would expect to find almost in an organ loft itself, where beat frequencies are obvious, to an en chamade Trumpet, slightly less loud.

The en chamade is loud enough to produce beat frequencies, but less so, and using C-C# separation, a major third comes from the same speaker but 5ths and minor 3rds come through separate speakers. The resultant beat frequencies are at the same level, but moving one's head causes movement through standing waves. Putting the stop through one speaker rather than separate speakers makes no real difference.

At lower sound levels, the diapasons suffer less. I have compared a mono diapason to a C-C# diapason with similar results and I will try to post a YouTube video of this later on. This should be interesting as using a camera with poor audio quality the effects should be more objectionable and it will also be interesting in its normalisation of signals on account of audio automatic volume control.

Best wishes

David P
http://www.organmatters.co.uk
David Pinnegar, B.Sc., A.R.C.S.
Offline
User avatar

toplayer2

Member

  • Posts: 1071
  • Joined: Mon Jul 23, 2007 2:15 pm
  • Location: Michigan, USA

Re: Beyond Stereo to multiple Speakers

PostFri Oct 01, 2010 6:56 pm

Another quick experiment. The file linked below was made using pure sine waves, the C above middle C is at 523.251 Hz and the E above it is tuned using Just Intonation to 654.063 Hz. This should rule out beats as a factor.

In the first five seconds, the two notes are mixed at equal levels into both the left and right channels. Then the notes are segregated with the C in the left channel and the E in the right channel. The overall level is exactly -20dB on both channels for both segments.

When I listen with headphones, the first segment sounds decidedly unpleasant to me while the second segment sounds pure. Other people report that it just sounds like mono turning into stereo with no unpleasantness in either segment. I am genuinely curious what is going on here.

https://sites.google.com/site/vtheatreorgan/Home/files/MajorThirdJustIntonationMixedThenLeftRight.wav

Joe
Offline
User avatar

David Pinnegar

Member

  • Posts: 437
  • Joined: Thu Jan 03, 2008 7:13 pm
  • Location: Sussex UK

Re: Beyond Stereo to multiple Speakers

PostFri Oct 01, 2010 8:03 pm

toplayer2 wrote:C above middle C is at 523.251 Hz and the E above it is tuned using Just Intonation to 654.063 Hz. This should rule out beats as a factor. When I listen with headphones, the first segment sounds decidedly unpleasant to me while the second segment sounds pure. Other people report that it just sounds like mono turning into stereo with no unpleasantness in either segment. I am genuinely curious what is going on here.
https://sites.google.com/site/vtheatreorgan/Home/files/MajorThirdJustIntonationMixedThenLeftRight.wav


Hi!

On my computer speakers both segments sound the same except for some reason the second louder than the first, no doubt my computer being slightly odd but importantly there is the beat note, presumably at 131Hz audible in proportion just the same in both.

I have now done the video promised demonstrating that there is effectively no difference in the level of beat notes whether the two notes come from one speaker together or individual speakers one for each note: http://www.youtube.com/watch?v=ntPblldKpBw

Incidentally, whilst the boxes visible and which have lent themselves to immediate comparison of the routing of these sounds to speakers, the whole instrument does not rely exclusively on them, four other types of sound generation being in operation.

Best wishes,

David P

(By the way, as a physicist, I don't like the modern fashion of simply reading unnecessary decimal places off digital displays, just because the digits are available. In the old days, we'd dial up a frequency of about 650Hz and and another at about 520 Hz on our sig-gens and the audible results would not be a great deal different . . . . The irrelevant and inaudible accuracy just gives us a false sense of thinking that we are in control and know what we're doing. I mention this not as a criticism but merely as a perspective in passing ).

Off topic, http://www.youtube.com/watch?v=_qqWNNlwV2U is an instrument I enjoyed exploring recently and I'll be posting more of this organ in due course
http://www.organmatters.co.uk
David Pinnegar, B.Sc., A.R.C.S.
Offline

jwillans

Member

  • Posts: 424
  • Joined: Thu Jun 16, 2005 12:30 pm
  • Location: Menston, UK

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 2:31 am

Hi Joe,

First many thanks for doing these experiments.

toplayer2 wrote:When I listen with headphones, the first segment sounds decidedly unpleasant to me while the second segment sounds pure. Other people report that it just sounds like mono turning into stereo with no unpleasantness in either segment. I am genuinely curious what is going on here.


To my ears the first is not unpleasant but does have a much slower beat and possibly slightly less separation of the waves. The second sounds to have a much more rapid and pronounced beat. Have you visually compared the wave forms to identify artificial artefacts?

James
Offline
User avatar

David Pinnegar

Member

  • Posts: 437
  • Joined: Thu Jan 03, 2008 7:13 pm
  • Location: Sussex UK

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 5:22 am

Hi!

Following the experiment on
http://www.youtube.com/watch?v=ntPblldKpBw
last night it occurs to me that in my experiment whilst two notes coming through one speaker were no different to the individual notes coming through individual speakers we all know that in practice big electronic organs with multiple speakers sound better. Why?

Pondering on this I wonder if it is a matter of dynamic range. In the experiments, all the speakers were being driven in safe regions of linearity. But when we put together an "organ", we vastly exceed just two "pipes", and whilst one added to another might be fine but another 80 and then another 150 sounding on top all at once may be a different matter not merely of sheer complexity but simply of dynamic range within linearity.

Could putting the whole of an organ sound of many thousands of pipes at realistic volume through a cylindrical hole perhaps 8 inches in diameter, or two such holes, upset our assumptions about the linearity of air compressivity?

The following file
http://www.jungleboffin.com/mp4/organ/h ... beware.mp3
may be capable of testing the dynamic range of speakers if played at a realistic volume.

Please take care: this file is speaker dynamite and may destroy inadequate speakers at a realistic volume and, most certainly, if it does not cause you to jump out of your skin or induce a heart attack, your speakers may simply not be hi-fi.

(The recording was made at our first concert to bring familiarity of the organ to new audiences, before I had heard of Hauptwerk on the unaltered, unimproved virgin 3 manual ex-Londonderry Cathedral instrument. No doubt it may illustrate how Hauptwerk is superior. Nevertheless I hope that you enjoy the fun as the audience did. The project to put the organ repertoire onto this concert platform owes a deep debt to the great organists who have enjoined with enthusiasm to make this possible, and thanks to Hugh Potton on this occasion. All organists are welcome)

Reverting to the concept of realism, should not the area of speaker cones in use be related to the total area of all speaking pipe diameters? In areas of speaker design requiring slots or reductions of openings from the cone, from memory the general rule is that a compression ratio exceeding 4:1 introduces distortion from non-linear air compression. So this might be a good point from which to work in relating organs to speakers.

Of course this is irrelevant to home reproduction systems where air shifting required to produce a "realistic" sound pressure is on a different scale to that required to be done by an instrument in its native building.

On this basis, taking a 1:1 pipe area to speaker area ratio an 8 inch unit might be capable of representing a dozen gamba and dulciana pipes, a tenor Diapason note played with a triad above middle C, but not more, one pedal Ophicleide, a myriad of mixture pipes, etc. One simply needs as many speakers as necessary to equate with the areas of speaking pipes, although physics can provide ways of defying this law.

In a smaller environment one can probably take the ratio up to 4:1 without undue effects, but in a church or auditorium greater than 1:1 is overkill.

This may be a reason in addition to my voicing observations recently why electronic organs have failed to measure up on full organ tuttis.

Best wishes

David P
http://www.organmatters.co.uk
David Pinnegar, B.Sc., A.R.C.S.
Offline
User avatar

mdyde

Moderator

  • Posts: 15441
  • Joined: Fri Mar 14, 2003 1:19 pm
  • Location: UK

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 7:15 am

Another quick experiment. The file linked below was made using pure sine waves, the C above middle C is at 523.251 Hz and the E above it is tuned using Just Intonation to 654.063 Hz. This should rule out beats as a factor.

In the first five seconds, the two notes are mixed at equal levels into both the left and right channels. Then the notes are segregated with the C in the left channel and the E in the right channel. The overall level is exactly -20dB on both channels for both segments.

When I listen with headphones, the first segment sounds decidedly unpleasant to me while the second segment sounds pure. Other people report that it just sounds like mono turning into stereo with no unpleasantness in either segment. I am genuinely curious what is going on here.

https://sites.google.com/site/vtheatreo ... tRight.wav


Hello Joe,

The apparent difference with headphones is just because the brain receives and processes each ear's signal separately, without mixing the signals from the two ears together. Both parts of your recording have the same frequency content, but you don't notice the unpleasant beating with the second part when listening on headphones. If you play your demo through a pair of stereo speakers that are close together and some distance from you (so that each of your ears hears the outputs from each of the speakers approximately equally) then both parts of your demo should largely the same, including the unpleasant beating.

I think that you've been confusing speaker intermodulation distortion with the 'subjective tones' psycho-acoustic phenomenon: if two frequencies are mixed together (whether in-air with speakers, through headphones, digitally, or whatever) and the difference between those two frequencies would be within the audible range, then your brain will be deceived into thinking it can hear a difference frequency, even though it isn't present in reality (which you can easily verify by recording the mixed sound with a microphone and looking at the recording's true frequency content using FFT):

http://hyperphysics.phy-astr.gsu.edu/hbase/sound/subton.html#c1
http://hyperphysics.phy-astr.gsu.edu/hbase/sound/beat.html

It's the same phenomenon 'acoustic bass' organ stops rely on, i.e. they trick the brain into believing that a 16' tone is present when it isn't in reality.

If a speaker manufacturer's IMD (SMTP) figure is about -69 dB (HR824MK2), then I really don't think IMD really is going to be noticeable or objectionable. Any IM distortion tones introduced by such a speaker should be *extremely* quiet, and almost certainly inaudible above the signal. Hence I don't think that IMD needs to be a significant consideration (for multi-channel audio or otherwise), provided that good-quality speakers are used and that they're not driven too hard.

I suspect that the main reasons that you don't notice the 'subjective tones' phenomenon (i.e. beats) so much with multi-channel audio are that:

1. Your brain is able to identify the sounds it hears from each of the speakers as separate (because of the spatial cues resulting from the differences in the patterns of reflections within the listening room resulting from the different speaker positions). Hence although the same frequencies are present in reality, the brain isn't so easily tricked into thinking that the beats are themselves tones.

2. The room reflections effectively randomise the phases of the various frequencies each speaker is producing at each point in the listening room ('signal mixing effect'). Because of the shifted phases, the beats may be heard more or less strongly at different points in space, so being able to move your head around (even slightly) allows your brain to separate the beats from the true frequencies more easily, again preventing it being so easily tricked into thinking that the beats are themselves tones.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.
Offline
User avatar

toplayer2

Member

  • Posts: 1071
  • Joined: Mon Jul 23, 2007 2:15 pm
  • Location: Michigan, USA

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 8:34 am

Martin,

You are probably offering the most plausible explanation, that this is a psycho-acoustic effect rather than IMD. From the last test, different listeners hear different amounts of the mystery tone. And as you suggested, playing through separate channels does not eliminate the effect, although the positioning of the two speakers can reduce it.

Last night I experimented again using the Tuba Horn. First, I zeroed out any tuning related beats by adjusting the E by -13.7 cents using Hauptwerk voicing. The mystery tone was still prominent. Then I forced the C to sound exclusively from the left speaker and the the E from the right using the stereo positioning control. The tone was diminished but still present. As you mentioned, slight head movements affected the intensity of the tone, presumably due to comb filtering.

A member of the VTPO forum suggested that the tones may be the product of "heterodyning". Wiki:

"Heterodyning is the generation of new frequencies by mixing (multiplying), two oscillating waveforms. It is useful for placing information of interest into a useful frequency range following modulation or prior to demodulation. The two frequencies are mixed in a vacuum tube, transistor, diode, or other signal processing device. Mixing two frequencies creates two new frequencies, according to the properties of the sine function; one is the sum of the two frequencies mixed, the other is their difference. These new frequencies are called heterodynes."

Again, this explanation has some appeal as well. Whatever the cause, I personally find the effect unpleasant and wish to minimize it insofar as possible.

Thanks to all contributors for a most interesting and enlightening discussion!

Joe
Offline
User avatar

mdyde

Moderator

  • Posts: 15441
  • Joined: Fri Mar 14, 2003 1:19 pm
  • Location: UK

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 9:05 am

Hello Joe,

A member of the VTPO forum suggested that the tones may be the product of "heterodyning". Wiki:

"Heterodyning is the generation of new frequencies by mixing (multiplying), two oscillating waveforms. It is useful for placing information of interest into a useful frequency range following modulation or prior to demodulation. The two frequencies are mixed in a vacuum tube, transistor, diode, or other signal processing device. Mixing two frequencies creates two new frequencies, according to the properties of the sine function; one is the sum of the two frequencies mixed, the other is their difference. These new frequencies are called heterodynes."

Again, this explanation has some appeal as well. Whatever the cause, I personally find the effect unpleasant and wish to minimize it insofar as possible.


Multiplying (as opposed to adding) frequencies together is something that should never happen in air, amplifiers, speakers, or anything else that processes audio. Audio signals always need to be mixed (in air or digitally) by adding, not multiplying.

Doing anything to a signal that would introduce new frequencies would be non-linear (not a linear time-invariant system), so speaker and amplifier manufacturers wouldn't do it (intentionally).

You'd also very easily be able to verify that that wasn't the case by recording the sound with a microphone and looking at its FFT (as I did in the screenshots I posted previously from your first IMD test recording). If the mysterious frequencies you think you can hear aren't visible (at an audible level) in the FFT, then they aren't there in reality, so they *must* be due to a psycho-acoustic (or in-ear) effect.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.
Offline
User avatar

TheOrganDoc

Member

  • Posts: 800
  • Joined: Sun Sep 19, 2004 10:13 pm
  • Location: South East, Florida, USA

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 12:05 pm

Hi all,

It appears to me that we need to approach someone with a properly tuned Pipe Organ
to carefully check this out. :roll:

The most common pipechest's, alternated pipes to each side of the chest, (ie. Leftside: C-D-E-F#-G#-A#)
so that places "C & E" just a few inches apart and facing the same direction !

I don't know if this helps but it is quite interesting, as in tuning a pipe organ,
the temperament is set by setting the proper "slow" beats alternating between fourth's and fifth's,
leaving the thirds and sixths (in equal temperament) notorious for their slight dissonance . :?

A past friend, and professional organist once quoted to me,
"that one should play the organ" and "listen carefully to the final result".

He then would say "never analyze the idiosyncrasy's of an instrument"
"or you will surely loose the enjoyment of it" ! :lol:

I will check this out on my setup shortly,

Best wishes, Enjoy !
Mel..............TheOrganDoc...............
Offline
User avatar

mdyde

Moderator

  • Posts: 15441
  • Joined: Fri Mar 14, 2003 1:19 pm
  • Location: UK

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 1:59 pm

Hello again Joe,

I recommend you try the following experiment to help you determine whether the tones you're hearing are really present (IMD) or not ('subjective tones'/beats):

Create a stereo WAV file with a few seconds of each of the following, in order:

1. A sine wave at 261.6 Hz (=middle C) in the left channel only.
2. A sine wave at 329.6 Hz (=middle E) in the left channel only.
3. Both the of the last two sine waves mixed (digitally) together, both at their original amplitudes and still in in the left channel only.
4. A sine wave at 261.6 Hz (=middle C) in the left channel only.
5. A sine wave at 329.6 Hz (=middle E) in the right channel only.
6. Both the of the last two sine waves mixed (digitally) together, both at their original amplitudes, i.e. with the C in the left channel and E in the right channel.

Now set up your Zoom mic/recorder and play the above stereo WAV file through a pair of your Mackie monitors, using the Zoom to record the result.

Assuming that you hear the mystery tone(s) when playing the WAV file, then use an FFT to examine each of the sections 1-6 from the recording you made with your Zoom. Maybe upload the result so that we can examine too if needed.

Any perceived difference tone would be expected at 329.6 - 261.6 = 68 Hz.

In particular, in the FFT you're looking for the frequencies present and their amplitudes in the recording when section 3 of the WAV file was playing (especially at 68 Hz). If any frequencies other than 261.6 and 329.6 Hz are present at significant amplitudes (relative to the amplitudes of the two test sine signals) then you've potentially identified IMD. If not, then you're hearing 'subjective tones' (beats) or some other psycho-acoustic (or in-ear) effect.

If you didn't hear the mystery tone(s), try repeating the whole experiment, but this time doubling all of the sine waves' frequencies (659.2 and 523.2) in the WAV file. (Be wary of just using an audio editor's pitch shift function on the whole file, since that could potentially itself introduce distortions.)

This time any perceived difference tone would be expected at 659.2 - 523.2 = 136 Hz.

If you still don't hear the tone, try a third time, doubling the frequencies again, so that any perceived difference tone would be expected at 272 Hz. Perhaps it's the case that your brain is more easily deceived into hearing subjective tones within a particular frequency range.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.
Offline
User avatar

toplayer2

Member

  • Posts: 1071
  • Joined: Mon Jul 23, 2007 2:15 pm
  • Location: Michigan, USA

Re: Beyond Stereo to multiple Speakers

PostSat Oct 02, 2010 3:18 pm

Martin,

Thank you for your suggestions. I will do my best to set aside some time to perform these tests, although my partner is rightly concerned about my diverting inordinate amounts of time away from needed work on the Paramount. One presumes that you have suggested equally tempered thirds for a reason.

I must confess that I have further confused myself by re-reading Colin Pykett's treatise on IM:

http://www.pykett.org.uk/EndOfPipeOrgan.htm#Intermodulation

I will also try to duplicate his findings and produce spectral analyses that support or refute them.

Thanks once again for sharing your very intelligent insights.

Joe
PreviousNext

Return to Amplification

Who is online

Users browsing this forum: No registered users and 5 guests