Soundstage and imaging

[Duke]

There is no such thing as this Audiophool thing called "soundstage".

What term(s) would you use when talking about spatial quality?

 

[valve_integrator]

There is no such thing as "spatial quality."
You are listening to transducers exciting the acoustic of your room.
Multi mic recording - used by the vast majority of recording systems/studios just makes even more artificiality.
Adding convolution is yet another kludge followed by my pet hatred - compression.
Artificial electronic stimulation produced by artificial recording techniques leads to a "postcard" or a "bathroom".

Read Michael Gerzon on spatialisation and coincidence in recording.

Read about multi channel + height generation info and dispersion from SPAT
- Ircam 1, place Igor-Stravinsky 75004 Paris

Convolution and computing Angelo Farina
Full professor of Environmental Technical Physics at the University of Parma.
Main research activity: Acoustics.
In more details: concert halls, musical instruments, subjective preference, auralization, numerical models for large rooms (pyramid tracing), small cavities (finite elements) and outdoors (image sources). Advanced measurement techniques including sound intensity, MLS, modal analysis, digital recording. And recently also DSP implementations, Underwater Acoustics, stretched-pulse measurements

The rest of discussion is a waste of space.
As usual on here.

SPAT

 

[valve_integrator]

I wouldn't call the best solution for most people most of the time primitive.

I certainly would, inc more than 90% of all recordings made.

 

[MattHooper]

LOL.

 

[Keith_W]

There is no such thing as "spatial quality."

Would you prefer the terms "apparent stage width" (ASW) and "phantom centre", or do you deny that they exist as well?

 

[STC]

That is a nice video. Thank you.

AT 9:24 he combines and ITD and ILD, and explains that we most likely will be hearing his sound coming from the Right channel as ITD wins over ILD generally. For me I was hearing it coming from Left channel. Tried it with 3 headphones, with EQ on and off - all of them still coming from the Left. Anyone had the same effect?

Same here. Only the left channel. I play a lot with ILD and ITD for crosstalk cancellation with loudspeakers and in real world both works with loudspeakers but generally ILD wins. I think he was trying to demonstrate HAAS effect but I think 15 dB is too high for the precedent effect to take place.

 

[valve_integrator]

Would you prefer the terms "apparent stage width" (ASW) and "phantom centre", or do you deny that they exist as well?

"apparent stage width" (ASW) and "phantom centre"??

I don't believe in this.
Being as I constructed my own system with 2-3 main criteria, I very much doubt such terms for a proper monitoring system are even close to valid.

The main criteria were:- (FYI)

1/ Almost completely flat response very little attenuation down to 25hz -*from 120hz down...(almost unknown)
Reproduces Albert Hall Organ quite well.

2/ Very careful adjustment and alignment to give an unusually wide neutral listening area. FRONT chs see 2b
(none of this "sweet spot" crap.)
2a/ Front large vertical offset with horn HF drivers - corrected with DSP to prevent driver resonance.
LF units up to 1m lower (physically)than the HF/Mid. (LF is mostly non directional)
2b/ Omni rear surround speakers with wide range drivers 1m+ above listener - leading to wide rear non directional ambience.

3/ Multi channel recording LIVE or using stereo signals to feed F +R systems at correct surround ratio.

"apparent stage width" (ASW) and "phantom centre"?
NONE OF ABOVE.
The so called stage width doesn't exist- it fills the room.
I don't use a centre channel.

To me this "sound stage" nonsense comes from poor room nodes and cancellation/subtraction at multiple frequencies.
(we don't need / want directional info being beamed at a listener)
+
a) The ability to confuse people with masking effects is clearly proveable.*
The inability of easily 70-80% of listeners to distinguish at blind test between high rate MP3 (with masking artefacts) and real high bit depth PCM from proper live material.. (Shocking but true)
The "Brainwash effect" of people constantly listening to processed digital music in the last 2 decades.

b/ Conditioning responses.

c/ Comb filtering of FR in people that have been conditioned this way.
eg. heard of "autotune"? Fake voices, fake intonation.

I remain highly dubious of any of the so called "high end" systems I ever auditioned.
These were usually in extremely sub optimal listening environments.
They failed convince me ever, that the whole thing was anything other than a big pile of BS.

Why do your own thing this way?
If you can't get anyone else to do a job properly - DO IT YOURSELF.

sub woofer tech is a can of worms, and handling LF [properly & cleanly, with good integration) requires maximum RMS power, + proper treatment.

NB.
Headphones (cans!) were no better - imaging behind the ear.
You may not agree, but plenty research evidence on this.

Perhaps some of the best recordings done were binaural with dummy head. (Bruel & Kjaer mics)
Make of that what you can.

 

[Keith_W]

"apparent stage width" (ASW) and "phantom centre"??

I don't believe in this.

Oh well. I got those terms from Toole's book. He believes it exists. I can also hear it (although I can't measure it), so I believe it exists. Not only that, but software like BACCH can dramatically expand apparent stage width and more precisely locate the phantom centre. I don't understand what processing they are doing, only that it has something to do with crosstalk cancellation, but it definitely works. I would be more than happy to take part in a blind test to prove that I can hear a difference in ASW with BACCH on and off if you are willing to put some money on the table

 

[MattHooper]

"apparent stage width" (ASW) and "phantom centre"??

I don't believe in this.
Being as I constructed my own system with 2-3 main criteria, I very much doubt such terms for a proper monitoring system are even close to valid.

The main criteria were:- (FYI)

1/ Almost completely flat response very little attenuation down to 25hz -*from 120hz down...(almost unknown)
Reproduces Albert Hall Organ quite well.

2/ Very careful adjustment and alignment to give an unusually wide neutral listening area. FRONT chs see 2b
(none of this "sweet spot" crap.)
2a/ Front large vertical offset with horn HF drivers - corrected with DSP to prevent driver resonance.
LF units up to 1m lower (physically)than the HF/Mid. (LF is mostly non directional)
2b/ Omni rear surround speakers with wide range drivers 1m+ above listener - leading to wide rear non directional ambience.

3/ Multi channel recording LIVE or using stereo signals to feed F +R systems at correct surround ratio.

"apparent stage width" (ASW) and "phantom centre"?
NONE OF ABOVE.
The so called stage width doesn't exist- it fills the room.
I don't use a centre channel.

To me this "sound stage" nonsense comes from poor room nodes and cancellation/subtraction at multiple frequencies.
(we don't need / want directional info being beamed at a listener)
+
a) The ability to confuse people with masking effects is clearly proveable.*
The inability of easily 70-80% of listeners to distinguish at blind test between high rate MP3 (with masking artefacts) and real high bit depth PCM from proper live material.. (Shocking but true)
The "Brainwash effect" of people constantly listening to processed digital music in the last 2 decades.

b/ Conditioning responses.

c/ Comb filtering of FR in people that have been conditioned this way.
eg. heard of "autotune"? Fake voices, fake intonation.

I remain highly dubious of any of the so called "high end" systems I ever auditioned.
These were usually in extremely sub optimal listening environments.
They failed convince me ever, that the whole thing was anything other than a big pile of BS.

Why do your own thing this way?
If you can't get anyone else to do a job properly - DO IT YOURSELF.

sub woofer tech is a can of worms, and handling LF [properly & cleanly, with good integration) requires maximum RMS power, + proper treatment.

NB.
Headphones (cans!) were no better - imaging behind the ear.
You may not agree, but plenty research evidence on this.

Perhaps some of the best recordings done were binaural with dummy head. (Bruel & Kjaer mics)
Make of that what you can.

I wouldn’t even know how to converse with this.

 

[Duke]

There is no such thing as "spatial quality."
You are listening to transducers exciting the acoustic of your room.

So what is your wording for the perception of spatial attributes such as different voices and instruments coming from different locations in space?

And what is your wording for the sensation of being within an acoustic space which is much larger than your playback room and which can vary greatly from one recording to the next? The attribute of "the sense of acoustic space is much larger than the playback room and varies greatly from one recording to the next" would be an indication that the spatial cues on the recording, rather than the spatial cues of the playback room, are perceptually dominant.

The so called stage width doesn't exist- it fills the room.

Is "the stage width fills the room" not a description of "the soundstage", and in particular a description of "the soundstage width"? And is "the stage width fills the room" not a "spatial quality"?

I'm trying to figure out whether you reject all perceptions and descriptions of spatial attributes in reproduced music, or whether you simply reject a particular vocabulary (and if the latter, then I'd like to know what vocabulary you use).

 

[Duke]

2/ Very careful adjustment and alignment to give an unusually wide neutral listening area. FRONT chs see 2b
(none of this "sweet spot" crap.)

Can you elaborate on what you are doing and how it results in "an usually wide neutral listening area"? (I'm not challenging your claim; I'm interested in learning what you are doing.)

2a/ Front large vertical offset with horn HF drivers - corrected with DSP to prevent driver resonance.
LF units up to 1m lower (physically)than the HF/Mid. (LF is mostly non directional)

So if I understand correctly, your horns are up fairly high and the low frequency section is very close to the floor? I can think of at least one good reason do to that. I would be interested in any details about the configuration, radiation patterns, crossover frequencies, whatever, that you are willing to share.

 

[valve_integrator]

I don't understand what processing they are doing, only that it has something to do with crosstalk cancellation, but it definitely works.

We do this on the OTHER end. (occasionally)
a single point mic system which does stereo recording magnificently well- via transaural rejection. (French idea)

It consists of a pair of omnis and a pair of cardios at the same source point.
The 2 cardios can be spaced a little apart at X-Y as per usual, but directly behind is a pair of omnis reversed in both phase and channel. It's something I never ever saw anywhere else (Primus SXB).

One engineer from another country asked me WTF is this thing - how does it work....ha!
The only answer I gave was shut up and watch - then we have to place it exactly the right height and distance.
The local radio people were also baffled opting for their crappy Decca triangle which didn't work of course....

On the reproduction end it struck me if you weren't following the same cues in reverse all was lost...
that is - same as recording, if the placing of the playback system is not exactly right for the chose space - you can forget it.

I once remember doing gaspard de la nuit in a crappy TV studio of FR3.
(recording a Steinway model D).
If you follow Michael Gerzon's approach, it does appear he was dead on the money, and a result constantly criticised the BBC (and I suppose their obessions with multi miking and mid side).

The result of the recording was fantastic, and a certain famous violinst (the mother of said pianist - A P ...S M) wanted to know how on earth I got that result in that studio -said she had never ever heard a recording like it.

Simple - like Gerzon says, there is an ideal way to listen - place mics and record.
He did most of this stuff in OXFORD.
Struck me playing it back is pretty much the same.

Look up Michael Gerzon.
He was like Blumlein, the closest thing I know to a genius, and British of course.

 

[STC]

Stereo can reproduce via loudspeakers playback the soundstage and imaging accurately but only under limited conditions.

1) If you make a recording of synth generated sax to the left speaker ( placed at 60 degrees triangle ) and synth generated guitar to the right speaker, you are essentially having two individuals playing in front of you.

2) However, recording of this sound and playing the same with the same speakers may not replicate the real event.

The recording and playback setup:-

R1) the microphone should capture the exact ITD and ITD.

R2) the reproduction should be with exact placement of speakers as the original event.

R3) room acoustics that can reproduce the reverbs.

R1 can be achieved with binaural recording. Better still if the dummy head and torso are the replica of the listener.

R2 and R3 can be achieved by playing the same in the room/venue it was recorded.

Yet it will not sound exactly as heard in live despite the chain of recording to reproduction using identical equipment.

The problems:-

P1) the sound captured by the microphone is not the same that is coming out from the speakers due to HRTF modification. This can sound accurate if played via headphones but not with loudspeakers as it requires correction filter to be applied. This is what Chesky’s binaural PLUS did. There are other problems with binaural headphones playback but I will skip that for now

P2) the original event produced was a single source event for each instrument. The reproduction is now via two speakers. The same sound is now played from two points in the recording. Is this accurate?

P3) the room‘s sound in the recording needs to be eliminated since it is being played in the same room otherwise this would now add more reverbs. So is the solution to play them in an anechoic chamber? Not exactly as we need lateral reflection for the sense of realism and stereo production can only produce the accurate stage within phantom stage. So realism suffers somewhat but for most having adapted to stereo soundstage after years of listening to them somehow learnt to overlook those errors. Those listening with headphones for ears not conscious or bothered about inside the head soundstage. We adapt to sound and learned to overlook the imperfection.

P2/3 is the inherent problem with stereo playback because it produces a duplicate sound delayed by about 220 μs to the wrong ear which is known as interaural crosstalk. So no matter what you do your soundstage width is confined to the maximum ITD difference position of the crosstalk. The reverbs too would now be confined by the two point from the speakers placement. This is very much different than in nature where the reverbs envelopes you from all direction. You lose more realism there. There is also the role of pinna for direction finding and it is extremely sensitive to the sound from 30 degrees angle. This is another weakness that would alter the exact imaging position in the phantom stage.

Crosstalk cancellation methods either using physical partition or DSP such as Apple spatial, Aria3D, RACE, BACCH somewhat eliminates the errors mentioned earlier so that sound extends beyond the physical position of the loudspeakers ( phase manipulated placements exempted) . The real need is for a soundstage that is more believable and natural rather than accurate reproduction of the soundstage. The goal of good stereo playback is to reproduce natural and live experience. Accurate reproduction via stereo is impossible without your HRTF used in the recording of the real event And that too will be limited to the frontal stage with limited lateral information.

 

[valve_integrator]

Stereo can reproduce via loudspeakers playback the soundstage and imaging accurately but only under limited conditions.

R1) the microphone should capture the exact ITD and ITD.

R2) the reproduction should be with exact placement of speakers as the original event.

R3) room acoustics that can reproduce the reverbs.

R1 can be achieved with binaural recording. Better still if the dummy head and torso are the replica of the listener.

The problems:-

P1) the sound captured by the microphone is not the same that is coming out from the speakers due to HRTF modification. This can sound accurate if played via headphones but not with loudspeakers as it requires correction filter to be applied.

sorry but this post is full of contradictions and an oxymoron...ie. hoping to do what Avery Fisher tried.
Confuse the audience and blind testing live v recorded music in his concert hall (ie. it's all been done properly before).

There are other problems with binaural headphones playback but I will skip that for now
So realism suffers s.... stereo soundstage after years of listening to them somehow learnt to overlook those errors. Those listening with headphones for ears not conscious or bothered about inside the head soundstage.

why ignore Binaural and dummy head recording?
The BBC tried it LIVE on Radio 3 years ago. (with of course mixed results) same as the results broadcasting
The Tempest by shakespeare in Matrix H surround btw....

We adapt to sound and learned to overlook the imperfection.

P2/3 is the inherent problem with stereo playback because it produces a duplicate sound delayed by about 220 μs to the wrong ear which is known as interaural crosstalk. So no matter what you do your soundstage width is confined to the maximum ITD difference position of the crosstalk. The real need is for a soundstage that is more believable and natural rather than accurate reproduction of the soundstage. The goal of good stereo playback is to reproduce natural and live experience. Accurate reproduction via stereo is impossible without your HRTF used in the recording of the real event And that too will be limited to the frontal stage with limited lateral information.

I already said there is no such thing as "soundstage".

You just proved the exact reverse of most of the points you set out to discuss, by comparing headphones and binaural with reality (they are NOT REAL, and has nothing to do with HRTF).

They are merely a different representation of reality from speakers.

Headphones are worse than hopeless for audio editing, no matter the price, but most broadcast engineers are using precisely this failed listening strategy, and hoping the final result is presentable.*

How do you expect to have listenable recordings this way when everything in the entire chain is either fake or artificial??
It's an invention of all those blingy audiophool marketing and glossy magazines.
The same people that claim they can hear extra details by changing speaker wire!

*One of my references is Janszen.
His father invented the ESL....Walker just copied some of the ideas.

 

[STC]

I already said there is no such thing as "soundstage".

Just because you said so it doesn’t stop to exist. Almost every discussion about stereo refers to the two terms.

I guess you are trying to say something important but I can’t figure out what it supposed to be.

sorry but this post is full of contradictions and an oxymoron...ie. hoping to do what Avery Fisher tried.
Confuse the audience and blind testing live v recorded music in his concert hall (ie. it's all been done properly before).

why ignore Binaural and dummy head recording?
The BBC tried it LIVE on Radio 3 years ago. (with of course mixed results) same as the results broadcasting
The Tempest by shakespeare in Matrix H surround btw....


I already said there is no such thing as "soundstage".

You just proved the exact reverse of most of the points you set out to discuss, by comparing headphones and binaural with reality (they are NOT REAL, and has nothing to do with HRTF).

They are merely a different representation of reality from speakers.

Headphones are worse than hopeless for audio editing, no matter the price, but most broadcast engineers are using precisely this failed listening strategy, and hoping the final result is presentable.*

How do you expect to have listenable recordings this way when everything in the entire chain is either fake or artificial??
It's an invention of all those blingy audiophool marketing and glossy magazines.
The same people that claim they can hear extra details by changing speaker wire!

*One of my references is Janszen.
His father invented the ESL....Walker just copied some of the ideas.

 

[Killingbeans]

I guess you are trying to say something important but I can’t figure out what it supposed to be.

It's like he wants something undefined to be undefined differently. All I get from it is a headache.

 

[audiofooled]

Just because you said so it doesn’t stop to exist. Almost every discussion about stereo refers to the two terms.

I guess you are trying to say something important but I can’t figure out what it supposed to be.

It is even possible to argue the meaning of existence. But in this case I would guess that the term of "soundstage" is something adopted by the audiophiles to describe the imaginary 3D space created by 2 channel stereo within listening environment. Originally the term is used to describe a stage or a set which is suitable for recording sound, especially in the filmmaking industry, so these are two entirely different concepts.

Of course there are other well defined terms such as sound field which is describing the environment in which the sound waves are propagating. For other well defined terms, such as ASW and LEV I highly recommend Dr. Toole's book.

I think what @valve_integrator is saying really is much more important than the argument about the term of "soundstage". Michael Gerzon was indeed a brilliant man and there are some of his papers archived here.

There are different techniques to extract more of what 2 channel stereo is, but at a cost of many adverse effects because of what 2 channel stereo truly isn't, as discovered by Michael Gerzon himself, and others, myself included.

Speaking only for myself as a DIYer, from personal experiments by using more channels and a DSP amplifier, the very positive side to this is the listener envelopment (LEV) and the illusion of space created by the phase and amplitude relations between front and rear speakers, also in room reflections. When listening to a 2 channel recording of a live event, this gives the more realistic sense of the space, also very detailed representation of recorded side effects of audience doing their own thing. It's as if I am there.

On the negative side, if you really want to hear all the flaws of stereo, especially at frequencies above 7kHz, alongside all the comb filtering and the timbral mismatch, which originally wouldn't be a problem, this is how you do it. There are ways to do it better, but it requires more elaborate setup.

In a nutshell, I got varying results with different recordings, largely depending on recording technique, which I couldn't do anything about. With some recordings there was great and life like performance, but with majority of others I ended up disconnecting more and more channels until only the main L and R are left, which sounded best...

In the end, I get the best overall results with only 2 channels, but with DIY loudspeakers and setup which is to some extent in accordance with research done by Earl Geddes. He says we need flat DI from 700 to 7000Hz, with dispersion characteristics no wider than 60 degrees, with a toe in angle of setup along which you get flat DI along the listening axis and not too much energy reflected to the sidewalls.

To me this method ensures that I do not hear any problems which are doomed to arise at higher frequencies, also reaching preferable direct to reverberant field, neutral timbre and adequate image precision with vast majority of the recordings. Low frequencies are very important, so do what you can with your subs. For me, stereo is better left uncomplicated, but that's just me.

 

[valve_integrator]

It's like he wants something undefined to be undefined differently. All I get from it is a headache.

Quite possibly because you have never been involved in the entire chain from Live recording to Broadcast audio + validation of a reproduction system.

Funny you still don't seem interested in Michael Gerzon's work.
Have you ever heard of him?
I don't think so.
Do I have to SPELL IT OUT?

"
Michael Gerzon (1945 – 1996) was an extraordinary mathematician, inventor, sound recordist, music lover and poet, who died prematurely from a life-long struggle with severe asthma.

Described in his obituaries as: “one of the audio industry's greatest thinkers and writers” , a “prolific polymath” and “genius”; he authored over 120 papers on aspects of spatial audio recording and reproduction, signal processing, systems theory and noise shaping.
However, perhaps his greatest achievements are the theoretical development of the Ambisonics surround sound system and, with Dr Peter Craven, the Soundfield microphone - together giving the ability to record and reproduce a three-dimensional 360˚ sound field from a single point."

 

[valve_integrator]

It is even possible to argue the meaning of existence. But in this case I would guess that the term of "soundstage" is something adopted by the audiophiles to describe the imaginary 3D space created by 2 channel stereo within listening environment.

Thanks for your excellent post which did better than I can to explain the issues.

I suspect I get something like this, as I took a lot of trouble measuring my system..... which is why I opted for (Russian made) horns quite similar to the highly regarded Altec 288G**. (but by Lomo SPB).
They are apparently rated to make over 104dB with 1W, but I had to tame them a bit & one of the diaphrams wasn't centred properly when I got it.
I tested them flat to 15.5khz with a litle null there, then took off again to 16.5khz.
The Alnico magnets weigh a TON! Never drop one on a toe!
fyi**288G are the last of the large format AlNiCo drivers

ie.
"dispersion characteristics no wider than 60 degrees, with a toe in angle of setup along which you get flat DI along the listening axis and not too much energy reflected to the sidewalls."

It took ages to get this system right.
Tolerances are to the order of mere cm, and +/- a few degrees on the mid ranges and the horns.
I was suprised just how difficult it was to set up and calibrate correctly, then measure using my Neumanns and Calrecs parallelled to cancel out any effects. (one is wide omni, the other cardio, both with known parameters)

"Low frequencies are very important, so do what you can with your subs"...

I don't have subs I have 1 x12" and 1x 15" in a giant sized cab x 2 channels (235L x 2) to handle anything down to 18hz, with deliberately no rolloff at all (nearly all systems have a 10dB per octave drop below 100hz).
And yes the entire front system is handled by 2 whopping great vintage valve amps. from 1963-64.

The acquistion sound card also has to be a known quantity.
(we found the same sound card behaved differently when run under Linux 32 compared with win32 - ouch!)

 

[audiofooled]

Thanks for your excellent post which did better than I can to explain the issues.

I suspect I get something like this, as I took a lot of trouble measuring my system..... which is why I opted for (Russian made) horns quite similar to the highly regarded 288G. (Lomo)

"dispersion characteristics no wider than 60 degrees, with a toe in angle of setup along which you get flat DI along the listening axis and not too much energy reflected to the sidewalls."

It took ages to get this system right.
Tolerances are to the order of mere cm, and +/- a few degrees on the mid ranges and the horns.
I was suprised just how difficult it was to set up and calibrate correctly, then measure using my Neumanns and Calrecs parallelled to cancel out any effects. (one is wide omni, the other cardio, both with known parameters)

"Low frequencies are very important, so do what you can with your subs"...

I don't have subs I have 1 x12" and 1x 15" in a giant sized cab x 2 channels (235L x 2) to handle anything down to 18hz, with deliberately no rolloff at all (nearly all systems have a 10dB per octave drop below 100hz).
And yes the entire front system is handled by 2 whopping great vintage valve amps. from 1963-64.

The acquistion sound card also has to be a known quantity.
(we found the same sound card behaved differently when run under Linux 32 compared with win32 - ouch!)

If the term, no matter how well or not so well defined is, communicates the same meaning to everyone within a single thread, then I don't see why not we can all play along.

DIY is not for everyone, but it can be very rewarding. The results are what matters, also the journey itself.

Setup can be critical depending on situation. In room, at different locations, there can be any number of phase relations between displacement, acceleration, particle velocity and pressure. Different measurement points give different results and multiple are required to give some information about gradient. What microphones hear is different than what our ears do.

My sealed sub is also a bit unconventional when it comes to application, it's basically a bass module going all the way up to 200Hz, so it took a fair bit of work to make it well behaved.