Understanding and Optimizing Tactile Feedback

[Kvalsvoll]

It's a tough call to me whether A or B has the better step response.  I prefer to look at the impulse responses instead of step responses.  IMO, your frequency response looks a bit better for A than for B.  I also think the waterfall for A looks better than it does for B.  I know from experience that even subtle improvements to the waterfall can be audible, but of course, it may not be the only factor at work here.

..

By the way, don't underestimate bass traps.  They made a huge difference in the enjoyment of bass in my space.  Even though the traps primarily work at > 80 Hz, they improve the response of the harmonics that give the lower bass hits and notes speed and definition.  A lot of tactile feel is also involved at these higher frequencies.  They also tend to make those lower bass notes sound louder.  Just looking at your responses (-10 dB from 95-140 Hz or so), I have no doubt that regardless of what your sound field looks like, you are missing out on a lot of bass that is very tactile.

 

The purpose here was to find differences in experienced sound when only sound field properties other than spl is changed.

The two configurations have differences in freq response and decay, we can see that.

I also believe the differences are large enough to be audible.

The B response can be improved to look more like A, but it takes some time and effort, and the switching between the two would take more time.

Since the initial impressions more or less matches my experiences with the original set-up, I chose to leave it like this, assuming it is close enough.

 

The hole around 100hz is lying a bit, if you move the mic the response changes.

However, it is very true that the measurements reveal poor control of decay and reflections from around 100hz and up into the midrange.

This can be fixed by adding absorption on the boundary planes that causes these reflections.

Because this room it set up to show something similar to a typical living room/media room, the amount and type of absorption is very limited.

Most of the problems around 100hz is caused by the ceiling, and it is also the ceiling that adds high-q resonances at and below 20hz.

(Of course I did not see this before taking the room into use, to fix it the ceiling must be taken down..)

 

What the measurement does not show is that the multi-sub approach improves the bass response from something like +-30dB with huuuge resonances to something that is completely smooth and with very controlled decay.

 

In this system the overall quality of bass improves when the subwoofer system is engaged, and it improves for all content, not just Oblivion with bass-eq.

You often see people talking about turning off the sub when playing music, because it sounds better without sub.

Well, it does not, if the subwoofer system is properly set up.

Also, all this is before we even start to consider any tactile feedback and velocity of the sound field.

 

Measurements - especially ones taken from one mic positions - will show deviations that are not necessarily true.

Moving the mic or changing the boundary conditions close to the mic will change the response considerably.

For bass frequencies the measurements presented here should be good enough, the mic is always kept in one position, and the most critical here is to be able to compare the A and B in the bass range.

[Kvalsvoll]

 

...

My own believe for why house curves sound better is that smaller rooms typically present transient sounds at higher sound levels than the larger rooms that content is typically mixed in.  This necessitates a lower overall playback level and a house curve to restore the tonal balance lost due to the lower playback level.

 

Unfortunately, I can't contribute my own experiments on this topic because I have a suspended wood floor that transmits vibrations quite well.  I also have a wide open room and may have bass conditions more similar to far field.  As it is, I could probably do with a bit less vibration in the upper bass range, where it seem my sofa and/or floor take a bit longer to decay than the sound does.

 

Amount of house curve, and shape, can be personal preference?

One thing I have noticed, though, is that the better systems tend to be preferred with a flatter response, the +20dB boost in the low end is abandoned.

 

The 3dB/decade bruel&kjær tilt also seems to be accepted, but I think it is important to notice here that the amount of tilt depends on the room. 

I tend to believe that if you place decent speakers with flat on-axis in a decent room, the resulting response is something similar to this tilt.

Note that the main speakers in A-B does not have any eq at all, they are born like that, and the room adds more and more as freq goes down, due to room acoustics and the speakers radiation pattern.

 

I don't think playback level is as important as suggested by the loudness curves.

With a good set-up I find that flat bass is both present and good sounding at low levels.

That the room is small should not require a lower spl, if it does, something with the sound field is different.

A small room with reasonable control of decay and reflections should be similar to a larger.

Movies at 0dB sounds great in the A-B room.

 

The A/B room has a wooden floor, this is what gives excellent tactile response below 20hz. 

If possible, I would prefer to not have any vibrations from sofa/seating and so at higher frequencies, it does not add any good to the experience.

The tactile feel from the sound directly is very different from feeling the vibrations from seating or other hard surfaces.

[SME]

Amount of house curve, and shape, can be personal preference?

One thing I have noticed, though, is that the better systems tend to be preferred with a flatter response, the +20dB boost in the low end is abandoned.

Of course it can be personal preference.  Or it can also be an honest effort to achieve a subjective experience that matches what the content creator and engineering team intended.  I would say a +20 dB low end boost is probably more of a preference thing in the first place.

 

The 3dB/decade bruel&kjær tilt also seems to be accepted, but I think it is important to notice here that the amount of tilt depends on the room. 

I tend to believe that if you place decent speakers with flat on-axis in a decent room, the resulting response is something similar to this tilt.

Note that the main speakers in A-B does not have any eq at all, they are born like that, and the room adds more and more as freq goes down, due to room acoustics and the speakers radiation pattern.

 

It definitely depends on both the room (and not just room size) but also speakers and placements.  Worse still, it depends on the content.  This is the problem of translation, which is widely known among professionals who mix and master audio.  The problem arises because systems are calibrated solely against a smoothed representation of steady state frequency response.  This is not enough to account for the level of direct sound (where applicable) and the level and overall pattern of reflections.  On top of that, we have to consider the time response of the ear, a subject on which there is limited data.

 

How speakers behave in a room depends a lot on the room.  Likewise, on-axis response is woefully insufficient to characterize a speaker's response in any given room.  One set of speakers that sounds good in one room may sound terrible in another.

 

 

I don't think playback level is as important as suggested by the loudness curves.

With a good set-up I find that flat bass is both present and good sounding at low levels.

That the room is small should not require a lower spl, if it does, something with the sound field is different.

A small room with reasonable control of decay and reflections should be similar to a larger.

Movies at 0dB sounds great in the A-B room.

Playback level doesn't dictate whether or not you can hear the bass, but it does affect the perceived balance as well as the tendency for content to disappear due to masking.  Deep and infrasonic bass are impacted the most.

 

Sound fields absolutely do differ between rooms and speaker configurations.  But I think it's generally more helpful to focus on the frequency response with different windowing intervals.  Compared to larger rooms, small rooms usually have stronger early reflections, faster reverberant decay, and less reverberant energy in general.  The strong early reflections enhance the loudness of the direct sound, and the overall reduction in reverberant energy makes the direct sound stronger than with a larger room.  This causes dynamics to be exaggerated and to sound louder than they usually do in much larger rooms like the dub stages used for final mix decisions for theatrical tracks.  Room treatments can help diminish early reflections but actually further reduce decay time and overall reverberant energy.  Hence, it's basically impossible to treat a small room to sound like a large one.

 

As for playing movies at "0 dB" in the A/B room, is that the case for all movies?  I would also point out that "sounds great" and "too loud" aren't exclusive.  In any case, I used to share your opinion on this, and I often blamed the movie/mixing for being too loud.  My views have changed from my experiences and from giving proper consideration to those who actually produce audio for a living.  The currently accepted practice seems to be to mix in the room of the same kind that you intend for playback, if possible.  Otherwise, adjust playback level based on room size.

 

I hope some day, the phenomenon of "translation" is better understood so that we calibrate our rooms on more objective rather than subjective measures of performance.  Until then, the best bet may be to just do it by ear.  One experiment worth trying on yourself is to download this:  http://www.atsc.org/refs/a85/Speech_sample.wav  Extract the audio from the left channel and play it in the center.  Close your eyes and adjust the volume purely by subjective loudness judgement and see where you end up.  The speech is standardized to -24 LKFS, which

is relatively loud for human speech but is not yelling.  Think of listening to someone address an audience.  The speaker is not usually yelling, but he/she isn't really using his/her "inside voice" either.  In principle, I shouldn't need to tell you this because our brains understand how speech sounds like so well that we can insinuate those things.  We don't need high SPL to tell when someone is yelling, yet we can easily identify whether any speech is being presented at a natural level or not.  A more typical level for dialog in movies is about -27 to -29 LKFS.  On my system, I usually blindly play this back at around -6 to -9 relative to a 85 dBC calibrated "0".  In other words, the most natural sounding calibration level for my room and system is 76-79 dBC.  This is consistent with the playback levels I use for movies and is also consistent with recommendations for small rooms and near field systems, and going by my measurements, my response is pretty close to near-field from top to bottom.

[SME]

The purpose here was to find differences in experienced sound when only sound field properties other than spl is changed.

The two configurations have differences in freq response and decay, we can see that.

I also believe the differences are large enough to be audible.

The B response can be improved to look more like A, but it takes some time and effort, and the switching between the two would take more time.

Since the initial impressions more or less matches my experiences with the original set-up, I chose to leave it like this, assuming it is close enough.

I agree that it's probably not useful to try to make the response at B look like A.  I have to admit that I'm a lot more open to the idea of particle velocity field playing a role in tactile perception after carefully thinking it through.  However, I don't believe this proposition can be tested without ruling out all other paths of vibration transmission.  It's too easy to be fooled.

 

At a recent time in the past, my room also had a dip in the upper bass similar yours, and even now it remains the roughest part of my response.  When I decoupled my mid-bass woofers, hoping to clean up their sound, I was surprised to find the bass seriously lacking, just as you describe above.  My measurements confirmed an almost identical SPL response aside from an overall loss of 1 dB or so and a negligible phase change, yet even when I pushed the sub level up 6 dB, it sounded distant and weak.  My wife strongly indicated likewise, and I removed the feet after having used them for a bit over a week.  We were both immediately glad to have the old sound back.  I can't be sure, but I'm almost certain it had to do with helping us discern what was going on in the 100-200 Hz range.  I had been tricked into thinking I was hearing upper bass that I was actually primarily feeling.  So I basically experienced a similar transformation in bass to what you are reporting, not by changing placements or distances but by merely removing mechanical decoupling feet from subs that were already located close by.

 

Once I get my upper bass as tight as I'd like, I'll try the decoupling experiment again and see if I like the result more.

[Kvalsvoll]

As for playing movies at "0 dB" in the A/B room, is that the case for all movies?  I would also point out that "sounds great" and "too loud" aren't exclusive. 

 

 

Exactly - it can definitely be too loud and still sound "great".

 

But 0dB is more or less required to get the right tactile feel and impact.

On a good, dynamic mix that does not clip, this is usually a suitable level, movies tend to vary in spl between scenes, so you get a brake in between the action scenes.

This lower spl also makes the returning action better, because the ears now has had time to adapt to the lower spl.

 

Two things will be missing if played at lower volume - the upper bass impact, and the sense of very powerful low bass.

 

Music is a different story.

It just gets better the louder you play, and often end up playing at +6dB.

That is too loud.

[Kvalsvoll]

 

It's too easy to be fooled.

 

It is always easy to be fooled when trying to judge sound.

We know that.

 

The upper bass in this system already has that sharp hit, without sounding thick. 

It is not only about having the hardest chest impact, which it certainly has not, but getting a balanced overall sound.

When the bass is right, it has this kind of roller-coaster feel to it, like when you go over an edge and dive down. just on a much smaller scale, much quicker.

 

Tactile feedback from vibrations I think is very different from the experience you get from the sound waves hitting you body.

[dominguez1]

But 0dB is more or less required to get the right tactile feel and impact.

On a good, dynamic mix that does not clip, this is usually a suitable level, movies tend to vary in spl between scenes, so you get a brake in between the action scenes.

This lower spl also makes the returning action better, because the ears now has had time to adapt to the lower spl.

 

Two things will be missing if played at lower volume - the upper bass impact, and the sense of very powerful low bass.

 

I've found Audyssey's DynamicEQ to do a very good job at heightening the tactile response at lower levels. It's still not the same impact as 0db, but is certainly better than lower levels without DEQ.

[maxmercy]

DynEQ is the only thing I miss about Audyssey.  I measured what it does, but it does a lot.  It looks ahead 1/24 of a second and adjusts EQ based on volume level continuously.  It applies a base correction depending on MVL, and adds more midbass to any signal under 0dBFS.  Replicating the base correction based upon your MVL helps, but it is the midbass that is added that adds the 'punch'.  It starts adding around 250-300Hz when adding to base correction.

 

Here's the base correction:

 

 

JSS

[Kvalsvoll]

I've found Audyssey's DynamicEQ to do a very good job at heightening the tactile response at lower levels. It's still not the same impact as 0db, but is certainly better than lower levels without DEQ.

 

Good advice, works much better than simply turning up the subwoofer channel, and it has an offset adjust so you can tune the bass-lift to your taste.

 

Kind of waiting for you @dominguez1 to comment on all those measurements and findings.

Maybe, basically what we are saying is "No, I don't think the world is flat, I think it is shaped like a half-dome, because that's what it looked like when i went up the mountain today."

[SME]

DynEQ is the only thing I miss about Audyssey.  I measured what it does, but it does a lot.  It looks ahead 1/24 of a second and adjusts EQ based on volume level continuously.  It applies a base correction depending on MVL, and adds more midbass to any signal under 0dBFS.  Replicating the base correction based upon your MVL helps, but it is the midbass that is added that adds the 'punch'.  It starts adding around 250-300Hz when adding to base correction.

 

Here's the base correction:

 

 

JSS

Your plot here doesn't show the whole story does it?  I believe Dynamic EQ also boosts the very top end and may also attenuate around 3kHz somewhat.

 

Yes, I think Dynamic EQ sounds very good, but I think it tends to boost surround levels excessively, IMO.  For that reason, I eventually stopped using it for movie playback.  It works fine for music in stereo, and if using PL2x, it is possible to shift some of the excessive surround sound back to the front using the "depth" control, although this workaround is not ideal.  Since I play all my music back using a PC, I may eventually attempt to implement my own solution.

 

For movies, I have my system calibrated with a rise in the bottom end that looks pretty close to the -10 dB curve you show here.  This way I can playback at a more "room appropriate" level calibrated at 76-79 dB and not come up short in the bottom end.  And yes, boosting the sub level alone is not good enough.  Doing so tends to make things too heavy on the mid bass compared to the upper bass and bottom end.

 

I'm still "bottom heavy" for music with dynamic EQ, but that's easy to deal with using the reference offset feature.

[SME]

Tactile feedback from vibrations I think is very different from the experience you get from the sound waves hitting you body.

 

I do agree with this statement.  However, when you are experiencing both sensations at the same time, I believe it can be hard to discern the two.

[Kvalsvoll]

I do agree with this statement.  However, when you are experiencing both sensations at the same time, I believe it can be hard to discern the two.

 

That is a good point.

Let's say you try to compare two different situations, and the actual difference is vibration feedback from seating.

Could be very possible that the one with more vibration will be judged as more tactile, believing this is due to more tactile response from sound waves, when in reality there is only more vibration.

[dominguez1]

Kind of waiting for you @dominguez1 to comment on all those measurements and findings.

Maybe, basically what we are saying is "No, I don't think the world is flat, I think it is shaped like a half-dome, because that's what it looked like when i went up the mountain today."

@Kvalsvoll

 

Not sure what to say about your measurements and findings...other than...

 

OKVINTMFHOUSE! 

 

That translates into, you da' man. 

 

Love the world is flat analogy...LOL. The half dome is very funny...but everyone knows it's 3/4 dome...

 

On a serious note; share your secrets with constructing a PVL probe. I know you mention it's in published materials readily available, but having a member contribute based on experience is far greater learning than gleaning info from another source. Several have asked, and not sure if they got lost in the shuffle, or you choose not to get into the details?

 

If you could document/snap pics of how you were able to take PVL measurements, more folks would be apt to try it and post measurements themselves. In my research, there has been NO available PVL or SIL data on this subject (subwoofers)....ever. This is truly "the world is flat territory". The more that folks know how to take measurements themselves, the more the art will become a science.

 

I'd also love to see you thoughts on a step by step optimization procedure for tactile feedback.

 

Not sure how all the above translates to you...rest assured, your work on the subject using PVL measurements is IMO, the first of its kind.  Well done.

[SME]

Yeah.  I have to admit at this point I'm willing to give consideration to the PVL thing.  It makes me curious enough to want to experience a "small room setup" where the SPL response is made flat using multiple sub placements and EQ but the PVL in the listening area is very low.  I listen in a room that opens to the full house, so I may not have the same problem with insufficient PVL that others do.

[Shredhead]

Whacky curves max.  What did that sound like at -20dB with music? 

[Ricci]

In my research, there has been NO available PVL or SIL data on this subject (subwoofers)....ever. This is truly "the world is flat territory".

 

 

You can model this behavior in various programs and it has recognized names so someone has data on it, but other than some very math heavy, technical papers there doesn't seem to be much on what effects there may be subjectively in a small room.

[Kvalsvoll]

On a serious note; share your secrets with constructing a PVL probe. I know you mention it's in published materials readily available, but having a member contribute based on experience is far greater learning than gleaning info from another source. Several have asked, and not sure if they got lost in the shuffle, or you choose not to get into the details?

 

No secret, and I will gladly share it with you.

One problem is to find an easy way to make a calibration curve, so that predictable, comparable and meaningful measurements can be made.

 

Another issue is that it requires some knowledge to make use of the measurements, and to make the measurements.

 

All this can be solved.

 

What is more difficult to share is descriptions of things I make in my company, that would be under-cover, or hidden marketing, and I don't like that.

I also may want to limit some of the technical descriptions, especially when it comes to solutions, as I would not like to see one of the marketing-driven suppliers suddenly appear with a xxxx-echo now without the -echo due to making use of my technology.

 

Some of the things I have posted here is around 9 months old now, the A-B experiment is new, I did it to have a fresh and new experiment to report on here.

I think there is enough knowledge now to try to make a short post describing how to avoid the worst situations, and how to improve a lot from that.

 

@dominguez1, remember to wear you helmet cam so that we can see on-line when you go over the edge, in your attempt to prove the world is actually 3/4 dome..

[Kvalsvoll]

Yeah.  I have to admit at this point I'm willing to give consideration to the PVL thing.  It makes me curious enough to want to experience a "small room setup" where the SPL response is made flat using multiple sub placements and EQ but the PVL in the listening area is very low.  I listen in a room that opens to the full house, so I may not have the same problem with insufficient PVL that others do.

 

My test-room is also partially open to one side, I think I mentioned it.

This certainly has a very huge impact on the resulting sound field, but still it is not difficult to achieve the "bad" situation.

The direction you play in - like, which wall you choose as your front wall with the speakers - also matters, and more important when you try to fix it - having the opening at the back wall is a huge advantage.

[dominguez1]

No secret, and I will gladly share it with you.

 

...snip...

 

Some of the things I have posted here is around 9 months old now, the A-B experiment is new, I did it to have a fresh and new experiment to report on here.

I think there is enough knowledge now to try to make a short post describing how to avoid the worst situations, and how to improve a lot from that.

That would be great if you could share how you created the PVL probe. Pictures and description would be great insight.

 

It would also be very insightful for you to summarize how to optimize PVL and Sound Intensity as well.

 

As Bosso has mentioned, is there a product opportunity that could be specifically created to optimize soundfields; particular near field placement?

[dominguez1]

@okv, in your tests, how much of the tactile feedback was indirect; meaning through vibration from the couch or floor vs direct to your body?

 

In my experience, resonances in the floor or couch in the ULF range tremendously enhance the overall experience. Take that away, and a lot is lost. The pressure and weight is still there, and your clothes moving as if a fan were on are also prevalent, but the 'shake' is no longer present.

 

Having said that, is there any way a riser could be build to a certain 'tune' or resonant frequency? 

[Kvalsvoll]

That would be great if you could share how you created the PVL probe. Pictures and description would be great insight.

 

It would also be very insightful for you to summarize how to optimize PVL and Sound Intensity as well.

 

As Bosso has mentioned, is there a product opportunity that could be specifically created to optimize soundfields; particular near field placement?

 

I have been thinking about a quick write-up, which I could make available for everyone interested, but I found I wanted to find a way to easily calibrate it, so that people don't waste time in setting up something that does not have sufficient accuracy for the purpose.

 

A short-form summary post could be a good idea, I can make it, but give me a couple of days.

 

As for product opportunity, I don't see how any stand-alone, reasonable thing that fixes all this could be realized.

It's all a combination of room, acoustics, speakers and subwoofers, and set-up.

And that is before we even talk about how the sound field should be, to give the desired results.

[Kvalsvoll]

@okv, in your tests, how much of the tactile feedback was indirect; meaning through vibration from the couch or floor vs direct to your body?

 

In my experience, resonances in the floor or couch in the ULF range tremendously enhance the overall experience. Take that away, and a lot is lost. The pressure and weight is still there, and your clothes moving as if a fan were on are also prevalent, but the 'shake' is no longer present.

 

Having said that, is there any way a riser could be build to a certain 'tune' or resonant frequency? 

 

Yes, this is a very important point.

I believe it was @SME that pointed this out as well - how can you be sure what kind of tactile feedback you are observing.

 

In the A-B example test, this was rather clear, as the movement from the floor was not very different, and the feeling of sound as shock waves hitting the whole body is quite different in feeling.

The bad configuration did not give any shock-wave feeling.

 

At very low freq I have not been able to verify that the velocity can be sensed at all, and I can not even hear the pressure.

I tried this at 10Hz, I don't remember the spl, I believe it could be 110dB, it was not less than 100dB.

At 12Hz the floor moves, and that makes it audible (or rather, sensible?) from around 85dB.

 

A riser can certainly be tuned, the resonance freq will be determined by the mass on the riser and its suspension - softer suspension and more weight gives lower tune.

I believe a riser with a well damped suspension will work best, while trying to get the tuning as low in frequency as possible.

[Kvalsvoll]

The "Velocity Meter for Experts" guide is ready, it is a simple guide to get you started, and - be warned - it is what it says, requires some effort and at least being familiar with REW measurements.

 

I will not post this on an open forum, so those interested will have to pm, and I will send you the text.
All users are of course welcome to receive this.

[dominguez1]

Okv, what do you think it would take to get accurate pvl readings for 80hz and below? Larger driver for a probe?

[Kvalsvoll]

Okv, what do you think it would take to get accurate pvl readings for 80hz and below? Larger driver for a probe?

 

That is what the description I sent you is meant for.

If you mean accurate without using any calibration or correction curve, a larger driver will not be good if you try to measure something very close to the sound source.

Finding a stand that can hold the driver stable while measuring will improve the measurement significantly at low frequencies, say below 20hz.

 

Even with no correction, your measurements can still be of use because all measurements you do on your system will be comparable.