Robert Harley moderates a discussion on room acoustics,
equalization, and DSP-based room correction with
TAS Senior Writer Robert E. Greene, Peter Lyngdorf of TacT Audio,
and Art Noxon of Acoustic Sciences Corporation.
REPRINTED FROM tHE ABSOLUTE SOUND MAGAZINE, OCTOBER/NOVEMBER, 2004
Robert Harley: I
think everyone would agree that room acoustics dramatically affect
the quality of reproduced sound. Given how important room acoustics
are to sound quality, why have acoustics been generally overlooked
by audiophiles, when they’ll worry about things like the purity
of the rhodium plating on binding posts?
Art Noxon: Air is
free. Sound is free. Once generated, sound is free—from the
speaker to your ears. You can’t tax it or insure it. It’s
taken for granted, and there are no bells and whistles, and you
can’t plug it into the wall like you can a piece of equipment.
RH: You seem to be
saying that because air moving in a room is intangible, it’s
overlooked.
AN:
Absolutely. It’s something that’s taken for granted
and it’s assumed everything is fine, unless you stop to think
about it. Or if you’ve got great components that don’t
sound great and you’ve exhausted all other options, then you
may then consider that the room is at fault.
Robert E. Greene: I think another factor is that people don’t
realize how badly the room is behaving. They buy all the nice equipment,
set it up, and start to listen. They hear some problems, especially
in the lower frequencies, say, below 500Hz where the room really
begins to have a dramatic effect, but they assume that’s okay.
I’ve had the experience over and over of going to someone’s
house with some measuring equipment and showing them that there’s
a giant hole at 200Hz and a big boom at 80Hz, and they’re
shocked. They’ve just been assuming that was what the recordings
sounded like.
RH: They don’t
have a reference.
REG: They don’t
know that the distorted tonal balance—mainly too much bass—isn’t
in the recordings. I think it’s one reason why audiophiles
are obsessed with female vocals. It’s because they know what
the human voice sounds like in that range. That range is not terribly
affected by the room.
AN: One of the easiest
ways to hear the room’s effect is to compare the sound of
speakers in a room with a good set of headphones. Just play anything
over headphones and listen to the musical clarity. Forget about
imaging. Just listen to the musicality of what you’re hearing
over quality headphones. That’s an easy, cheap way to get
in contact with the phenomenon we’re talking about.
RH: Peter, you have
extensive experience with demonstrating DSP room-correction systems,
showing people at hi-fi shows what music can sound like with the
room-effects removed.
Peter Lyngdorf: I certainly have, and I’ve set up in a huge
number of rooms over the years. The fact of the matter is that below
roughly 300 or 400Hz, the room is the overwhelming factor in reproduced
sound, and below 300Hz is where you have about 90 percent of the
energy in an average piece of music. Below middle C on a piano,
the sound is totally controlled by the room.
Most of the tonality of music is destroyed
by an average room, and it is not only the tonality but the delays.
Every time you have a resonance in the room, you also have a delay,
which means that every time you extend the frequency response downwards
with larger speakers, you will hit lower and lower resonances. Every
one of those resonances is accompanied by a delay. If you have a
resonance at 30Hz of 10dB, you are actually accumulating ten times
the energy at that frequency, which means that the average delay
at 30Hz then would be about 0.6 seconds. Consequently, the energy
at that frequency comes almost at the next beat of the music.
So we have two issues: one is the
tonality, the other is the timing. And timing is totally out the
window if you do not have your room/speaker interface under control,
and I believe that’s one of the reasons that most musicians
seriously hate hi-fi. The music is so totally out of beat that they
can’t stand it. And the bigger the speaker and the more powerful
the system, the more they hate it.
Another thing which is quite interesting
and actually quite funny is that a lot of manufacturers are consistently
saying that their equipment needs to break in for two weeks or three
weeks or four weeks, and I believe that is almost entirely nonsense,
because very few products change dramatically over a few weeks.
I think what they wait for is that the customer runs through all
of his CDs one more time and finds another two or three tracks that
sound pretty good in his room.
RH: On the face of
it, equalizers appear to be a tempting solution to fixing these
problems, but audiophiles have abandoned them. Is equalization fundamentally
flawed, or has the problem been in the execution of the products?
REG:
Everybody in the consumer world has got it in his head that equalization
adds phase shift, but, of course, it’s really just the other
way around. If you equalize the resonances out of a room, the correct
timing is restored. The resonances themselves are generating phase
shift. The resonances are technically known as “minimum phase.”
When you take away the resonance with equalization, the timing—the
phase—also corrects itself.
AN: Equalization
has disappeared from high-end audio. All the image-distorting widgets,
which include equalizers, were taken out of high-end systems, just
about the same time that the [Dahlquist] DQ-10 appeared and speaker
designers realized that they had to get their speakers phase-aligned.
I’m surprised to see equalization back in high-end audio—if
it is back at all.
REG: I don’t
actually agree with that, at least in the bass. I think the phase
shifts in the bass are generated by the room and eliminated by the
EQ.
PL: Well, my view
is that equalizers work, but it’s important to have a good
room to start with. So I don’t think we are in total disagreement
with Art. It’s just that at the lower end of the frequency
response, where you have the most amount of energy, it’s extremely
difficult to remove those big fundamental resonances by acoustic
means alone.
REG: Even just a
few adjustable parametric equalizer coefficients will enable you
to get control of the room’s major disasters. It’s a
little on the crude side compared to the automated DSP systems,
which as you know have scores or sometimes hundreds of parametric
coefficients. But even analog equalizers with adjustable center
frequencies let you attenuate those two or three discrete frequencies
that are going “boom-boom.”
RH: That leads me
to the next topic, DSP room correction. Is DSP room correction a
panacea that solves the problems we’ve been talking about?
REG: My experience
with the DSP devices is that you can do amazingly effective things
with them. When set up right, it’s amazing how well they work.
RH: And what’s
also amazing is how dramatically different the sound is, when those
resonances are removed. It’s not a subtle effect.
PL:
It’s certainly something that anyone can hear. But some audiophiles
are scared about this kind of equalization because it implies that
you’re shifting bits and doing all kinds of things to the
signal, when everybody has been trying to keep the signal as pure
as possible. What a lot of audiophiles still do not realize is that
with digital technology you can do an awful lot that doesn’t
really change the musical information, add distortion, or create
noise as the old EQs did. Room correction can precisely hit the
room problems without adding distortion or noise. That, I think,
is why the early equalizers did not work—they could never
exactly hit the problem.
RH: Peter raised
an interesting point about audiophiles having an aversion to signal
processing. In the digital age, do we still want to keep the signal
path as simple as possible, or does digital technology open up new
frontiers that weren’t available in analog audio? [This topic
will be the subject of a future TAS Roundtable-RH]
PL: The goal here
is to make a piano sound like a piano and a cello like a cello,
with all the oomph and body you get from the real instrument. I’ve
heard so many high-end audiophile systems where the midbass is simply
removed. And then a lot of audiophiles are clapping their hands
because it is oh-sodetailed, and the female voice is beautiful,
and it’s clean. The only problem is that the instruments don’t
sound like actual instruments anymore. And that’s where the
combination of sensible room treatment with sophisticated room correction
can make the instruments sound like the real things. A piano is
one of the most difficult things to reproduce, and with the proper
EQ and proper setup of speakers you can really make a piano sound
like it is in front of you. And I don’t think that is possible
unless you use both a sensible room treatment and very sophisticated
DSP equalization.
RH: So we still need
acoustic treatments and correct loudspeaker placement—DSP
room correction isn’t the magic bullet.
PL: If you have the
chance to separate your low-frequency devices from your mid/high-frequency
devices [i.e., a subwoofer with small loudspeakers], you should
place your mid/high frequency devices where they give you the best
possible imaging and the bass units all the way into the corner
of the room. If the bass units are in the corner of the room the
whole initial sound from the woofer is minimum phase. It is all
going in the right direction at once. Whereas if you try to reproduce
the low frequencies from your main speaker standing freely in the
room, then much of the bass will go back into the corner of the
room, reverse, and then come dripping back to you in the next 4,
5, 6, 10 milliseconds.
AN: That was the
phenomenon the TubeTrap was designed to fix back in ’83. It’s
a bass trap placed in the corners behind the speaker with a treble-range
diffuser that faces forward into the room. There was no such thing
as a hi-fi subwoofer back then. With full-range speakers, the wavefront
would expand back into the corner and fold back out. Trapping that
bass with a TubeTrap put this company on the map and has kept us
alive for twenty years so far. Our approach is to absorb that bass
so it doesn’t get reflected back into the room, and to backscatter
the mids and highs with a polycylindrical diffuser. That rear bass
energy you’re talking about is the same thing we’ve
been addressing with Tube Traps for more than two decades.
REG: I’d like
to introduce a slightly heretical thought here. If you have a resonance
which you then remove by precise equalization, you’ve actually
made your speaker in some sense happier. It does not have to work
nearly as hard. The frequency range that had a 10dB boom has now
become a range where basically the speaker can produce the required
level with 10dB less output. Corner placement is actually nice for
speakers because it insures maximum coupling to the room and minimum
power input required to generate the bass you want. But, of course,
you have to add some time delay if you’re going to put the
woofers in the corners.
AN: When you build
a reverb chamber for acoustic testing, the classic speaker position
to stimulate all the room’s resonant modes is in the tri-corner
[the point where two walls and the ceiling or floor intersect].
This is a room made from concrete two-feet thick coated by two inches
of polished marble, and that has a reverberation time of fifteen
seconds. The classic position for the measurement microphone to
pick up all those resonances is also in a tri-corner.
The goal of speaker placement is to
avoid stimulating room resonances, so I’m pointing out an
inconsistency here with regard to corner placement of a subwoofer.
The typical position to avoid stimulating resonances is 29 percent
of the room’s dimension off the floor, 29 percent in from
the sidewall, and 29 percent in from the end wall. That is the most
neutral position possible. I was recently reading a 1974 AES paper
where the fellow produced a very smooth response in the room simply
by moving the subwoofer around.
But now we have the opposite proposal
of putting the subwoofer in the corner. It may have the benefits
discussed, but it stimulates all the resonances. We have two diametrically
opposed ideas. One is to put it at the minimum resonancestimulating
location in the room and achieve fairly flat response, and the other
is to put it in the corner and apply DSP room correction to correct
for all the distortions that are introduced by stimulating the room
modes. I think it’s important that we be clear about these
opposing perspectives.
PL: I agree they
are opposing, but try to look at it this way: corner placement of
the woofer will give you more total energy. Why will it give more
total energy with the same amount of excursion from the cone? Simply
because less of the energy generated by the speaker will be canceled
by out-of-phase components interfering with in-phase components.
So once you do precision equalization, you will get better timing
of the signal. Really, you can say that the more SPL [sound-pressure
level] you get at the listening position from a certain placement
of the woofer system the less the signal you are receiving is out
of phase and canceling. Of course, in most cases you end up getting
too much energy, but that is so easy to take away with good DSP
room correction. So that’s my very simple argument for corner
placement.
AN: If you have no
acoustic treatments in the room and you have a DSP processor, what
happens to the articulation in the room? You’re still injecting
energy into the room. You still have reverberation time. You still
have the lack of intelligibility that you had before. You still
have bass energy circulating in the room. People confuse DSP with
intelligibility. Intelligibility is the ability of a room to rapidly
respond to the dynamic changes of sound. We want a fast build-up
and a fast decay. DSP doesn’t address the decay rate factor
of rooms, and neither does equalization. There is no electronic
sound absorber for sale.
REG: None of us wants
to try to correct your marble reverberation chamber with DSP—in
the mids and highs. But in the bass, room ringing really can be
cancelled out, at least for one listening position. The adapted
DSP filter rings, too, but in reverse phase to the room ringing,
so the combined result is that ringing is gone.
RH: I’d like
each of you to comment on what you think is the single most important
thing that audiophiles can do to improve the sound of their rooms.
REG: My feeling is
damping the first reflections off the ceiling, floor, and sidewalls
makes a fantastic difference. I have a preference for speakers with
narrow radiation pattern, but that’s a long, complicated subject.
Most audio systems have way too much energy in the mid and higher
frequencies bouncing around the room and not nearly enough of the
first arrival. The first arrival is, among other things, where the
imaging information is.
AN: We do rooms all
day, every day, all over the world, for twenty years, and the first
reflections are not what we fix first. The first thing we attack
is something we call “head-end ringing.” We shorten
the vertical and lateral reverberation time in the bass by treating
the space to the side of, and behind, the speakers. It’s one
of our trade secrets. By cleaning up head-end ringing, we dramatically
expose low-level detail in the midrange and treble. I agree that
absorbing first reflections improves imaging, but without treating
head-end ringing, there’s this slug of vertical and lateral
shaking of air that oozes past the listener about a twentieth of
a second after the direct signal and blurs musical detail and imaging.
After we’ve addressed head-end ringing, then we control first
reflections.
PL:
I’ve done a lot of testing on the effects of reflections in
rooms, and there was a big, big project in Denmark about twelve
years ago, with a lot of companies involved in investigating effects
of reflections in rooms. I had the pleasure of being a test person,
where we could actually simulate the audible effect of the floor
reflection, sidewall reflection, ceiling reflection, and so on independently.
The single most disturbing reflection in the room is the floor reflection.
That is what makes the speaker sound like a radio and not like the
actual event. The second worse reflection is the ceiling reflection.
Sidewall reflections, if they are sufficiently delayed (more than
about five milliseconds) and are left/right symmetrical, can be
actually beneficial to the sound. But if your speakers are very
close to the sidewalls, you have to kill the side reflections. But
do not be too concerned about the sidewall reflections. The floor
reflection absolutely must be handled, followed by the ceiling reflection,
either by absorption or diffusion.
AN: Well, I’ve
been there, and diffusion blurs imaging. Ceiling diffusers blur
imaging, and they add an artifact because they’re a bunch
of resonators. All of our ceiling traps have reflectors in them,
but the reflectors are always offset and point to the back of the
room. Diffusers are energy storage devices, are tonal in nature,
and create incoherent reflections, which mask the perception of
coherent reflections.
PL: I’m sorry.
I didn’t express myself properly. I was not thinking about
diffusers as the devices you can buy. I’m thinking something
on the ceiling that will prevent the first reflection from reaching
you directly.
RH: I’d like
to conclude by adding my own comment about the single most effective
technique for improving the sound of your room: loudspeaker placement.
Through loudspeaker placement you can control the amount of bass,
overall tonal balance, image specificity, and soundstage width.
Speaker placement has its limitations, and the sound will still
be greatly influenced by the room, but I suspect that most readers’
systems could be improved by better speaker placement. ¦
AN:
Absolutely. It’s something that’s taken for granted
and it’s assumed everything is fine, unless you stop to think
about it. Or if you’ve got great components that don’t
sound great and you’ve exhausted all other options, then you
may then consider that the room is at fault.
REG:
Everybody in the consumer world has got it in his head that equalization
adds phase shift, but, of course, it’s really just the other
way around. If you equalize the resonances out of a room, the correct
timing is restored. The resonances themselves are generating phase
shift. The resonances are technically known as “minimum phase.”
When you take away the resonance with equalization, the timing—the
phase—also corrects itself.
PL:
It’s certainly something that anyone can hear. But some audiophiles
are scared about this kind of equalization because it implies that
you’re shifting bits and doing all kinds of things to the
signal, when everybody has been trying to keep the signal as pure
as possible. What a lot of audiophiles still do not realize is that
with digital technology you can do an awful lot that doesn’t
really change the musical information, add distortion, or create
noise as the old EQs did. Room correction can precisely hit the
room problems without adding distortion or noise. That, I think,
is why the early equalizers did not work—they could never
exactly hit the problem.
PL:
I’ve done a lot of testing on the effects of reflections in
rooms, and there was a big, big project in Denmark about twelve
years ago, with a lot of companies involved in investigating effects
of reflections in rooms. I had the pleasure of being a test person,
where we could actually simulate the audible effect of the floor
reflection, sidewall reflection, ceiling reflection, and so on independently.
The single most disturbing reflection in the room is the floor reflection.
That is what makes the speaker sound like a radio and not like the
actual event. The second worse reflection is the ceiling reflection.
Sidewall reflections, if they are sufficiently delayed (more than
about five milliseconds) and are left/right symmetrical, can be
actually beneficial to the sound. But if your speakers are very
close to the sidewalls, you have to kill the side reflections. But
do not be too concerned about the sidewall reflections. The floor
reflection absolutely must be handled, followed by the ceiling reflection,
either by absorption or diffusion.