Why might a voice or instrument move it's apparent location in the soundstage?

[5-pot-fan]

Hello,

I am a bit reluctant to post this, as I don't know if this is a basic beginners query or if it is a complex matter, but I have not yet found a direct answer to it, apart from the item in the thread referenced below*
By and large I have a soundstage that is acceptable, given the room size (3.9m x2.6m x 1.2m high) and the equipment (noted below). In most of my listening voices and instruments are satisfactorily separate in the soundstage but not pinpointed (although I do not know if they always should be).

However,listening to some opera I heard a couple of odd things, so here are 2 examples:
Ex 1. Nozze di Figaro – Act 1 scene 8 No 10 – Aria. For me the male voice (baritone) starts left of centre and moves to centre and back again as the power of the vocals increases and decreases. This also accompanies an apparent rise and fall in pitch.
Ex 2 Act 2scene 1 No 11 – La Contessa - definite movement from left of centre towards centre as pitch /volume increases, then back as these decrease.
The apparent distances moved seem unlikely to coincide with on-stage movements by the performers.
The next example did not behave the same:
Ex 3 - Act2, scene 2 No 13 Susanna's Aria – very little apparent movement.

I have also noticed a similar issue with electric guitar in at least 1 other audio track. Here my impression was that the instrument moved from right speaker some 10/20 degrees towards centre, then back again as loudness/frequency fluctuated. (I give this example to rule out a specific single speaker problem).

Based on my examples it seems possible that this effect happens at the change between tweeter and woofer in my speakers (Linn Katans,) as this area coincides with a large part of the human vocal range. If the crossover is the reason for the apparent soundstage issue, how could I test this? (I have REW+ Umik-1).

There are no manufacturers published figures for the Katan's crossovers.

Alternatively, is 'directivity' (or poor implementation of) the issue?

Does this apparent effect have a common name (that I could search on)?

If anyone can confirm my suspicions or offer alternative explanations, please do so!

My current set up is: Rpi4/Moodeaudio 9.1.3 → Denon X3600 → via preout to Linn AV5125 containing active crossovers for Linn Katans. MLP is 1.8mfrom each speaker, which are 1.75m apart and 0.6m from the front wall.

* thread in ASR -“Soundstage, and how much influence can audio equipment have”
Quote from DonH56 Aug10 2018 “And a speaker that does not integrate the drivers well will exhibit changes in soundstage and image as pitch goes up and down and so forth.”

 

[AnalogSteph]

given the room size (3.9m x2.6m x 1.2m high)

You have a room only 1.2 m (4') high?

Anywho, your speakers seem to have a 6.5" woofer and a 3/4" tweeter sans waveguide. I would be willing to go out on a limb and say that it's virtually impossible to get even directivity out of a combo like that, which in turn would tend to explain your wandering phantom sources. You can make 5" plus 1" work reasonably well, but things go downhill quickly beyond that. New speaker time, I'm afraid. Would you have any specific budget in mind? Probably in GBP?

 

[alex-z]

160mm woofer + 19mm tweeter will cause a large directivity mismatch in the crossover region. You cannot fix this with the crossover, you need physical changes to the speaker such as a waveguide for the tweeter.

3 way speakers are the optimal approach because you can keep a small mid-range for good directivity and still use a larger woofer for good bass output.

The sharp baffle edges are another issue, there will be significant peaks and dips caused by diffraction. Again, not an issue you can fix via the crossover or EQ because the on-axis and off-axis behaviour differs. A simple 19 or 25mm roundover on the cabinet edges will address most of this behaviour.

Room acoustics are the final hurdle. You have a relatively small room so your sound field will be dominated by early reflections. Adding broadband absorption to the first reflection points (side walls, ceiling, and rear wall) will allow you to hear more of the natural stereo imaging from the speakers. 3.5" mineral wool panels are a cheap and effective option, especially if you build them yourself.

 

[solderdude]

If its room acoustics listen to the same music on a headphone.
When the effect stays and others don't hear it then it is your brain/hearing.
When the effect goes away it is your room/speakers.
When the effect stays the same it is likely the recording or some processing.

 

[5-pot-fan]

Thank you all for your replies so far.
Correction to room measurements - height is 2.2m!
@solderdude - thank you - I should have thought about using headphones!
So, is it the case that poor directivity (however caused) is what is responsible for my apparent issue?
I understand that the crossover in the Katans would be difficult if not impossible to ameliorate, and that new speakers would be very nice to have. This must wait a little while, but when I start looking would the Revel PerformaF206 speakers be suitable in the existing room? Is their directivity considered sufficiently good/capable?

 

[solderdude]

Poor directivity/dispersion pattern (speaker cabinet design + drivers) could be the cause for a shifted stereo image for instance when different tones are played and the position in the image changes depending on the note.
It would not be listening level dependent just frequency dependent.

Solutions can be room conditioning (hard walls, floors), speaker placement/orientation, listening position change or using EQ with a measurement mic in the listening position.
Another solution is possibly using different speakers that have a different/better/more suited dispersion pattern.

 

[goat76]

However,listening to some opera I heard a couple of odd things, so here are 2 examples:
Ex 1. Nozze di Figaro – Act 1 scene 8 No 10 – Aria. For me the male voice (baritone) starts left of centre and moves to centre and back again as the power of the vocals increases and decreases. This also accompanies an apparent rise and fall in pitch.
Ex 2 Act 2scene 1 No 11 – La Contessa - definite movement from left of centre towards centre as pitch /volume increases, then back as these decrease.
The apparent distances moved seem unlikely to coincide with on-stage movements by the performers.
The next example did not behave the same:
Ex 3 - Act2, scene 2 No 13 Susanna's Aria – very little apparent movement.

Can you link to your 3 examples, it's nearly impossible to find specific tracks in classical music without at least seeing the album art.

 

[kemmler3D]

If the voice is moving during a crescendo on male voice, IMO it's probably because of directivity error in the general vicinity of 2khz.

Basically as the balance of frequencies in the voice shifts, so does the direction in which the speaker is throwing the most energy, and unevenly so.

This means the sound representing the voice will be (unless you room is perfectly symmetrical) bouncing toward you from a slightly different direction. In some cases this makes it seem like it's moving.

 

[DVDdoug]

I'd guess it's in the recording, either intentionally or unintentionally... Have you listened to these recordings on a different system?

Sometimes instruments are "actively panned" but if was done as an effect it would be more than 10 or 20 degrees.

If you can find a frequency-sweep file you could play it to see if it stays in the center. If it does stay in the center it's probably in the recordings.

"You never know what the problem is until it's fixed", so you may need to move your speakers to another room, or try different speakers, etc.

...And of course the soundstage is an illusion with the sound actually coming from a pair of speakers.

 

[Casper]

You haven't mentioned which recording you are listing to. It is entirely possible--especially if it is a live recording--that the singers are moving.

 

[5-pot-fan]

@goat76 The recording is from 1970, with Elisabeth Sodersrtom, Geraint Evans, Teresa Berganza, and Otto Klemperer conducting. Re-released on EMI CD 1990's I think. I no longer have the CDs. I will listen again over headphones in a day or so and report back.
From the replies above it seems that directivity may be the key, and I will do some searching on this topic.

 

[DonH56]

Reflections from boundaries can affect the frequency response and cause the position to vary as frequency goes up and down. Testing with headphones as @solderdude said is the usual test to see if it is room/speakers or the recording. These days, room correction can also mess with the image, particularly for different listening positions. A bad speaker could also cause the issue, though that is less likely, and usually causes more obvious problems.

 

[Pareto Pragmatic]

You haven't mentioned which recording you are listing to. It is entirely possible--especially if it is a live recording--that the singers are moving.

Or singing/playing in a different direction, given multiple mics being used in recording most live events. Turning a head, looking right to signal the sax player you are about to turn it over for their solo, things like that.

 

[kemmler3D]

I found this on the forum, it's not the exact same speaker but it does seem to have trouble in the upper vocal range. What do you want to bet there is also something funky with the directivity up there? https://audiosciencereview.com/foru...el-m106-vs-stock-linn-katan.23509/post-815642

If the crossover is as high as it (maybe) looks in that image then the woofer is undoubtedly beaming before then.

 

[goat76]

@goat76 The recording is from 1970, with Elisabeth Sodersrtom, Geraint Evans, Teresa Berganza, and Otto Klemperer conducting. Re-released on EMI CD 1990's I think. I no longer have the CDs. I will listen again over headphones in a day or so and report back.
From the replies above it seems that directivity may be the key, and I will do some searching on this topic.

If the below YT video is from the same album you are listening to, I don't think you have to worry too much about your speakers as the vocals are all over the place in the stereo field.

 

[Titurel]

@goat76 The recording is from 1970, with Elisabeth Sodersrtom, Geraint Evans, Teresa Berganza, and Otto Klemperer conducting. Re-released on EMI CD 1990's I think. I no longer have the CDs. I will listen again over headphones in a day or so and report back.
From the replies above it seems that directivity may be the key, and I will do some searching on this topic.

As this recording is of "The Marriage of Figaro," it's quite possible the singers were moving around. Some opera recordings in the early stereo era were staged - London's recordings notably. Considering how important movement is in "Figaro," I could easily imagine this recording was at least partially staged, even if it is slightly outside the time window when the practice was fairly common.

 

[5-pot-fan]

@solderdude
@DonH56

and others who have replied - thank you.
I have now listened on headphones and while the apparent movement in the soundstage is different, it is much less marked. I imagine that the headphones have their own characteristics, and having the signal so much closer to my head, will make some differences, but the fairly abrupt changes in apparent location when listening to loudspeakers are not there on headphones. So, it would seem that this is less of an issue with the recording and more to do with the speakers and/or the room. That makes it easier to research at least.

@kemmler3D
You offered 2 comments I would like to follow up on:
a) "Basically as the balance of frequencies in the voice shifts, so does the direction in which the speaker is throwing the most energy, and unevenly so."
b) "If the crossover is as high as it (maybe) looks in that image then the woofer is undoubtedly beaming before then."
How or why does a speaker driver throw energy unevenly, and what is meant by 'beaming'?
I have a mental picture of sound coming from a speaker cone, the direction being based on the cone size and shape (forget wave-guides for the moment!), but do different frequencies come from different parts of the cone, or is this an ignorant conjecture?
I think a 'Speaker Driver Course 101' would be helpful to me at this stage, and wonder if you or another contributor could point me towards any such material (definitely beginner level!)

 

[DonH56]

@solderdude
@DonH56

and others who have replied - thank you.
I have now listened on headphones and while the apparent movement in the soundstage is different, it is much less marked. I imagine that the headphones have their own characteristics, and having the signal so much closer to my head, will make some differences, but the fairly abrupt changes in apparent location when listening to loudspeakers are not there on headphones. So, it would seem that this is less of an issue with the recording and more to do with the speakers and/or the room. That makes it easier to research at least.

NP And, quite possibly, and/or a combination.

@kemmler3D
You offered 2 comments I would like to follow up on:
a) "Basically as the balance of frequencies in the voice shifts, so does the direction in which the speaker is throwing the most energy, and unevenly so."
b) "If the crossover is as high as it (maybe) looks in that image then the woofer is undoubtedly beaming before then."
How or why does a speaker driver throw energy unevenly, and what is meant by 'beaming'?
I have a mental picture of sound coming from a speaker cone, the direction being based on the cone size and shape (forget wave-guides for the moment!), but do different frequencies come from different parts of the cone, or is this an ignorant conjecture?
I think a 'Speaker Driver Course 101' would be helpful to me at this stage, and wonder if you or another contributor could point me towards any such material (definitely beginner level!)

This article does not delve into speaker directivity and such but might be helpful: https://www.audiosciencereview.com/forum/index.php?threads/comb-filter-effects.25432/

This one discusses room modes: https://www.audiosciencereview.com/forum/index.php?threads/room-modes.25433/

Beaming occurs when the sound field radiated by the speaker narrows. At very low frequencies, the sound wavelengths are very long, e.g. >11 feet at 100 Hz. At 1 kHz, they are about 1', and <1.5" at 10 kHz. When the wavelength is greater than the size of the driver, the speaker acts more like a point source, sending sound (radiating) in all directions. As the wavelength shortens and becomes comparable to the size of the driver, the sound pattern narrows, becoming more like a "beam" as the wavelength approaches the size of the driver. This is one reason smaller drivers are used for higher frequencies -- less beaming, e.g. broader radiation pattern.

This article shows wavelengths vs. frequency: https://www.audiosciencereview.com/forum/index.php?threads/wavelength-vs-frequency.1918/

HTH - Don

 

[kemmler3D]

How or why does a speaker driver throw energy unevenly, and what is meant by 'beaming'?

There are two reasons the energy might be dispersed unevenly at a given frequency.

Beaming is the common name given for destructive interference near a speaker cone that results in the sound being projected primarily forward, narrowly, from the speaker cone, as opposed to all directions.

The reason for this is as the sound waves get higher in frequency, they get smaller. As the wavelength approaches and becomes smaller than the size of the cone, the waves start to interfere with each other as they come off the cone.

If you imagine these 3 speaker icons are actually one big speaker, you might start to see how that works. Beaming is an inevitable consequence of using a speaker cone to play frequencies that are too high, i.e. smaller than the cone.

Sound waves also interfere if they come from the tweeter and the woofer at the same time. This is why designing a crossover is so tricky.

The second reason the energy might be dispersed unevenly is diffraction. This happens when a sound wave encounters something similar in size to itself, it bends around corners or discontinuities in the surface. In speakers this means the corners of the baffle or other features on the front of the speaker have some effect on the direction / distribution of the sound.

I have a mental picture of sound coming from a speaker cone, the direction being based on the cone size and shape (forget wave-guides for the moment!), but do different frequencies come from different parts of the cone, or is this an ignorant conjecture?

All the frequencies come from all parts of the cone, but if the waves are smaller than the cone, you can (sort of, kind of) think of the waves as individually coming from different parts of the cone and colliding with each each other in front of the cone. The net effect when the waves are much smaller than the cone is "beaming", i.e. most of the sound aside from the very center of the cone gets cancelled out.

 

[VientoB]

Headphones have way less soundstage than speakers.