Room correction limits

From DRC

Quote: "Room Correction is Impossible"

The answer below is from Denis Sbragion, the creator of the room correction software DRC. He was writing in response to an email from Lars Tofastrud. You can see the original message here:
http://www.avsforum.com/avs-vb/showthread.php?s=&threadid=445559&perpage=20&pagenumber=8

---

Being the developer of DRC I'd like to say otherwise, but for the most part what Lars say is true. Despite all the efforts placed in it DRC is able just to correct a minimum part of the room problems. For example take a look at this picture:

This is what gets corrected in the time/frequency plane for various values of one parameters of DRC, and pretty much defines the physical limit for electronic correction. The part corrected is the one below the curves, and the default value for the parameter is 1.0. As a rough extimation I would say that no more than 2% to 3% of the whole response gets corrected.

By luck our ear sees this time frequency/plane in a "distorted" logarithmic fashion. Following this view the picture above becomes this one:

Here things get better, and from our ear point of view I would say that about 20% to 30% gets corrected. BTW it is still no more than 1/3 of the actual room problems.

Some people have argued that for such short period of time even the time scale should be considered logarithimc. If this assumption is true the graph above becomes like this:

In this graph about 50% of the plane gets corrected, but I'm not convinced that the "early log perception" assumption is true, so I stick with "no more than 20-30%". BTW this graph is quite usefult to understand where the physical limits for room correction are placed and how my DRC program works.

After saying this I have to add some more to the topic. Even though so little can be corrected electronically it doesn't mean it isn't worth the effort. There are some thing to take into account:

• Electronic correction gives its best in the bass range, i.e. exactly where traditional passive correction is more problematic and require big elements that are difficult to place, expecially within normal listening rooms

• About in the same range where digital room correction is effective it is concentrated most of the musical signal, and this is also the range where our ear focuses while listening to music

• Electronic correction corrects also the direct sound, where passive correction has no effect at all

So it's clear that combining the two options (passive and active) is probably going to provide results that are really close to perfection even within normal listening rooms, where a complete passive correction is always difficult and sometimes impossible, for the quite obvious reasons that often the listening room is also a living room.

Apart from this I want to argue some of the Lars arguments specifically.

   quote:

   It's simple: it is not possible.

Well it's possible at least up to a limited extent. The important thing is to match expectation with results.

   quote:

   When the sound has left the loudspeaker it is too late and hence; to
   correct for what comes out of the loudspeaker is to introduce errors
   that loudspeaker designers do their best to weed out.

This is true only if you do magnitude only correction. If you correct also the phase up to a certain amount you can correct both the direct sound and the reverberant field. This amount is higly frequency dependant, as could be seen from the graphs above.

Another way to see this is to say that digital room correction corrects only the direct sound, but the concept of direct sound is frequency dependant. And after all, in the bass range, where's the line separating direct resume sound from the room effect? Basically we can say that in the bass range the speakers and the room are just one single thing, and you have to correct both at the same time, unless you have an enormous listening room.

   quote:

   For low frequencies you can correct for a few dB on the frequency
   response with a parametric EQ but the decay time of the mode will never
   change (although plenty people claims so). Apart from the very severe
   +-10dB deviations it is probably the decay time you perceive and not
   the frequency response as such.

Again, this is not true. If you equalize also phase you have also some control on the time behaviour. Not an enormous effect, not something that lets you put absorbers, bass traps and Helmoltz resonators in the trash bin, but indeed something that is worth the effort.

   quote:

   For comb filters (at higher frequencies) you need to lower the level of
   the reflection and there are simply no way to correct for comb filters
   in your listenign position electronically. At about -28dB (relative to
   the direct sound) the reflection will have no influence on the direct
   sound and what reaches your ear will be what comes out of the
   loudspeaker.

This is instead absolutely true. Just looking at the graphs above it's clear that above 1-2 Khz digital room correction is nothing more than speaker correction. If you want more you have to go the traditional way, but luckily absorbers are no more so problematic above those frequencies, and may be that playing a little with carpets, furniture and the usual things we are used to, we could get quite good results even outside of properly designed control rooms.

I hope this clarifies a little what digital room correction is, and what it isn't. All the discussion here is clearly focused on my DRC program, but the most part could be applied to digital room custom writing correction in general, 'cause it is tied to unavoidable physical limits that don't change when you change your filter design program.

Bye,

Denis Sbragion

---

To help understanding the reply above here it is the full text of the original message from Lars Tofastrud.

    I am not 100% sure if i am reading you correctly here but I am reading
    it that you want to correct for room anomalies with EQ.
    
    As a acoustician i have to say i do get a little frustrated when
    reading articles and even seeing products that supposedly correct for
    room problems. Why? It's simple: it is not possible.
    
    When the sound has left the loudspeaker it is too late and hence; to
    correct for what comes out of the loudspeaker is to introduce errors
    that loudspeaker designers do their best to weed out.
    
    For low frequencies you can correct for a few dB on the frequency
    response with a parametric EQ but the decay time of the mode will never
    change (although plenty people claims so). Apart from the very severe
    +-10dB deviations it is probably the decay time you perceive and not
    the frequency response as such.
    
    The only way to solve issues with the room is to deal with it in the
    room. This is done by inserting "resistors" to lower the Q of your
    room modes. When the Q value goes down the peak it self lowers and the
    modes get a wider frequency range. In theory you can never have too
    much bass trapping in a room. In reality too much absorption steals
    too much efficiency and small 10" and 12" sub-woofers bottom out trying
    to give you 80dB in the good seat. This is one of the reasons why you
    need to know a lot about the modes of your specific room to deal with
    them in a realistic way. Also, in the modal range of your room the
    standing waves will develop at specific places (directions) and placing
    a trap in the wrong place can almost be compared to putting a component
    in the wrong place in an electrical circuit: it will have no or an
    unwanted effect. In a properly treated room the soundstage (if there
    is such a thing in the recording) will extend way down in the bass area
    and it will not seem to come from all around.
    
    For comb filters (at higher frequencies) you need to lower the level of
    the reflection and there are simply no way to correct for comb filters
    in your listenign position electronically. At about -28dB (relative to
    the direct sound) the reflection will have no influence on the direct
    sound and what reaches your ear will be what comes out of the
    loudspeaker.
    
    Effectively you now have an anechoic room... Not very realistic or
    practical. That is why F. Alton Everest compiled the book "the master
    handbook of acoustics" and people like me work as studio designers :-)
    (this email is getting way to long...)
    
    regards
    Lars Tofastrud