Can Loudspeakers Accurately Reproduce The Sound Of Real Instruments...and Do You Care?

[RayDunzl]

I will test this in a spreadsheet and post the results.

Audacity easily produces what you want to see, and you can hear it.

F1 and F1*2 and the combination, first, "in phase", then, at some other phase angle.

The two mixed waves will sound the same, though they look quite different.

.

 

[MRC01]

...
Thanks Definitely interested here.

It looks like you posed FFTs, and I don't think that can show the effects of a comb filter. If you start with 2 waves, the FFT is only going to show those 2 waves. If you give one a phase offset, that will simply appear in the FFT. You won't see how their superposition affects amplitude unless you sum them!

 

[MRC01]

Audacity easily produces what you want to see, and you can hear it.

F1 and F1*2 and the combination, first, "in phase", then, at some other phase angle.

The two mixed waves will sound the same, though they look quite different.

.
View attachment 26475

Exactly. And there should be phase angles that do sound different from others. Put differently: as you slowly change the phase angle from 0 to 360, the sound of the output wave should change audibly. At least, my intuition says it should. Though intuition is sometimes wrong, so I'm checking it.

 

[RayDunzl]

And there should be phase angles that do sound different from others.

That tends to not be the case.

Try it.

 

[andreasmaaan]

It looks like you posed FFTs, and I don't think that can show the effects of a comb filter. If you start with 2 waves, the FFT is only going to show those 2 waves. If you give one a phase offset, that will simply appear in the FFT. You won't see how their superposition affects amplitude unless you sum them!

I thought you were saying that phase shift will affect the frequency (amplitude) response of the summed output?

And there should be phase angles that do sound different from others. Put differently: as you slowly change the phase angle from 0 to 360, the sound of the output wave should change audibly.

I agree that it might sound different, just not that the amplitude response will be different.

 

[MRC01]

... as you slowly change the phase angle from 0 to 360, the sound of the output wave should change audibly. ...

That tends to not be the case. Try it.

I am!
It's definitely true if the two frequencies are the same, or close to the same. Then, as you change the phase angle, the amplitude drops to zero then rises back up again.
I'm testing whether similar amplitude shifts happen if one frequency is an integer multiple of the other.

 

[RayDunzl]

Exactly. And there should be phase angles that do sound different from others. Put differently: as you slowly change the phase angle from 0 to 360, the sound of the output wave should change audibly. At least, my intuition says it should. Though intuition is sometimes wrong, so I'm checking it

Consider a piano.

Play an interval (two notes)

Does it sound different if you play the notes simultaneously, or one at some slight delay? (not talking about the attack, but the sustain).

Play them simultaneously, over and over. Your fingers won't precisely start the two notes at any defined phase angle.

Different sounds due to phase?

I think "not really".

And, to the point, the relative phase of the two notes could be anywhere.

Does the pianist worry about the phase among the strings of his big chord? I've not heard that discussion. What would he do about it? sub-millisecond timing of his fingerings?

 

[MRC01]

Here's the hole I see in that analogy: we know that if both F1 and F2 are the same, then there is no sound when they're 180* out of phase. Can somebody with 2 pianos hit the same note on each, and time them so that we hear nothing?
No he can't. Part of the reason is because each piano note has many different frequencies so it's impossible to find any timing that has them all 180* out of phase with the same note on a different piano.

Also, as musicians tune we hear beat frequencies when the pitches are close together, yet slightly off. This makes it easy to get perfectly in tune. Thus, 2 slightly different frequencies do indeed have aggregate amplitude warbling when superimposed. I believe this is similar to a comb filter effect. One frequency is so close to the other, when viewed in a short time scale window it looks like 2 of the same frequency slightly shifted in time/phase.

 

[MRC01]

...
I agree that it might sound different, just not that the amplitude response will be different.

If there are phase angles that sound different, then phase response can throw off timbre reproduction, whether or not it changes the amplitude response.
PS: that is a big "IF" and yet unproven. It's an interesting concept I'm playing around with.

 

[andreasmaaan]

If there are phase angles that sound different, then phase response can throw off timbre reproduction, whether or not it changes the amplitude response.
PS: that is a big "IF" and yet unproven. It's an interesting concept I'm playing around with.

I wish I could recall the source, but I do believe that it's been demonstrated that changes in phase angle can audibly affect the sound of two tones. It can certainly affect the ability of a masker tone to mask a maskee tone (that I can source: Zwicker & Fastl, Psychoacoustics), so it's reasonable to assume it must also be able to affect the perceived tonal balance of two tones.

It's the changed amplitude response I'm not convinced of, however...

 

[MRC01]

I wish I could recall the source, but I do believe that it's been demonstrated that changes in phase angle can audibly affect the sound of two tones. ...

So far, my spreadsheet shows only small changes in total output power as the phase angle between 2 frequencies changes. Enough to be audible (about 5% or half a DB), but I need more time to explore because:
1. I'm not sure whether these differences are real, or cumulative rounding errors.
2. Even if total output power were always the same, it wouldn't necessarily imply they all sound the same.
3. Output power seems to vary with phase angle for any 2 frequencies, and I haven't yet discovered anything to support the intuition about one being a multiple of the other.

What's interesting about the FT approach is that it's not fully invertible. Take some waves, sum them up, FT the sum, then build a new signal from the FT components. You don't always get what you started with. Example: if you start with 2 sin waves of equal freq and amplitude 180* out of phase, the sum signal is all zeroes, the FT of that signal is all zeroes, so the inverse FT is all zeroes. OK so that's a special case. Is it possible to do this when the waves you start with don't have the same frequency? Probably not because Fourier theory says the FT is unique for the smooth waveforms that make up audio signals.

 

[andreasmaaan]

So far, my spreadsheet shows only small changes in total output power as the phase angle between 2 frequencies changes. Enough to be audible (about 5% or half a DB), but I need more time to explore because:
1. I'm not sure whether these differences are real, or cumulative rounding errors.
2. Even if total output power were always the same, it wouldn't necessarily imply they all sound the same.
3. Output power seems to vary with phase angle for any 2 frequencies, and I haven't yet discovered anything to support the intuition about one being a multiple of the other.

What's interesting about the FT approach is that it's not fully invertible. Take some waves, sum them up, FT the sum, then build a new signal from the FT components. You don't always get what you started with. Example: if you start with 2 sin waves of equal freq and amplitude 180* out of phase, the sum signal is all zeroes, the FT of that signal is all zeroes, so the inverse FT is all zeroes. OK so that's a special case. Is it possible to do this when the waves you start with don't have the same frequency? Probably not because Fourier theory says the FT is unique for the smooth waveforms that make up audio signals.

Will be interested to see what you come up with

Keep in mind I said earlier, "I do believe that it's been demonstrated that changes in phase angle can audibly affect the sound of two tones."

In other words, agreement with your second point above.

OTOH, I'm not aware of any evidence that moderate changes in phase (of the type and magnitude generally present in conventional minimum phase crossovers) can have audible consequences with music or speech.

 

[Hifi]

Matt....l had asked ....

Question for you regarding your listening set up . . . What instrument would you say is the most difficult for your speakers to reproduce with realism?

I will open the question to all.

I will go first. Piano. Not just in my speakers and system but in about every system I listen to.

 

[Blumlein 88]

Matt....l had asked ....

Question for you regarding your listening set up . . . What instrument would you say is the most difficult for your speakers to reproduce with realism?

I will open the question to all.

I will go first. Piano. Not just in my speakers and system but in about every system I listen to.

The cannon in the 1812 overture.

 

[RayDunzl]

Banjo.

 

[Blumlein 88]

Banjo.

I would hazard a guess the banjo is a microphone selection and placement problem.

 

[Theo]

Matt....l had asked ....

Question for you regarding your listening set up . . . What instrument would you say is the most difficult for your speakers to reproduce with realism?

I will open the question to all.

I will go first. Piano. Not just in my speakers and system but in about every system I listen to.

I would think that a snare drum would be quite difficult to replicate...

 

[Cosmik]

What instrument would you say is the most difficult for your speakers to reproduce with realism?
...
I will go first. Piano. Not just in my speakers and system but in about every system I listen to.

What is your opinion of the digital pianos that proliferate? i.e. literally trying to replace an instrument with speakers?

 

[RayDunzl]

I would hazard a guess the banjo is a microphone selection and placement problem

Actually, I'm very satisfied with my playback.

And as you infer, the recordings are likely to be the weakest link in a competent playback chain.

Doesn't stop me from enjoying a great composition or the performance of it.

I won't listen to drivel for more than a moment just because "it sounds good".

In the end, I suppose I hear through the faults.

Put something on after something else, you can say "this sounds like crap!", then you adjust to it, and it sounds like what it sounds like.

To say "this piano doesn't sound like this other piano I think I have a memory of" is a bit self defeating, maybe.

*takes off a funky sounding Miles Davis - Live Evil, and puts on a funky sounding Hiromi Uehara - Place to Be...

Ah, it was the right choice, solo Hiromi. I kept waiting for the drums and bass to kick in, they didn't.

You need good power for a believable piano.

106.9dB peak, 83.7 average on track 2...

23.2db peak to average is on the high side...

14.45x average voltage, 208.92x average power (if the calculations are valid)

The amplifier heat sinks are 131 and 135F... in a 78F room.

Sounds like there's a Yamaha Concert Grand in here right now... And she's pounding it...

Somebody else's photogenic example of my old-school gear:

For Sale! Only 11.500,00 €

About 275W idle power draw, so, it ain't green...

Nearly antique Krell FPB 350 mcx monoblocks (the babies of the series) and Krell KCT preamp

MSRP was something like $12,500 for a pair and $8,500 for the pre in 2005 or so.

I got mine on eBay.

Sometimes you just have to throw caution to the wind and get what you think you want.

The flourescent display on the pre is a little dim, hasn't changed though over the 8 years I've had it. One of the blue LED indicator lights on the right amp was flaky for a couple of days last year. Other than that, faultless. Powerful. Clean. Dynamic. Noiseless.

Measured a possible 0.00035% distortion on the preamp yesterday (in another thread). Expect to try the amps tomorrow, just out of curiosity.

The noise on the left and the slope on the right are the Behringer UMC202HD's footprints.

Piano?

Yeah.

 

[oivavoi]

As a hobby recordist and musician myself, my best recordings come from putting my Rode NT1a mics in an ORTF like setup, then "balance with my feet" stepping the mic pole closer or further from the musicians to get the right voicing & balance. That can be as close as 6' or as far as 20' away depending on the size of the group and the room. The mics are spaced about the same as human ears. This gives a really stunning stereo image, nearly holographic 3-d.
Modern recordings made like this are rare. Yarlung and Mapleshade make some recordings like this, but they're the exceptions. Most are close-miced and mixed into artificial stereo. You can get more detail that way, and if tastefully done, the artificial stereo doesn't have to sound like a wall of sound, even if it's not as good as a true 2-mic recording. For detail and timbre, this can actually be superior, giving the freedom to match individual mics to the instruments.

I agree with this. What I would really like is some list of recordings done in this fashion - minimalist two-mic recordings, be it classical or jazz or other musical genres. I have searched for such a list in vain. Perhaps you'd like to volunteer to start such a list on a new thread here in the forum?