So, I'm just catching up here. Great discussion. Robin Bowes;228865 Wrote: > > Room can be thought of as a filter. e.g > > Signal-->Speaker-->Filter-->Ear > > So, if we want to remove the effect of the Filter(i.e. the room), we > need to apply the inverse of the Filter to the signal. e.g. > > Signal-->1/Filter-->Speaker-->Filter-->Ear > > If the filtering is done in the digital domain with sufficient > precision > and accuracy it should have little or effect on the signal.
If only things were that simple! The "room filter" is quite multidimensional - but the correction s really just one-dimension (all the sound comes out of the same loudspeakers). I think Darren catches some of that in the post I'm quoting below. To visualize why it's so hard, try a couple of scenarios: - There's a wall to the right of the loudspeaker, fairly close. So the sound at the listening position has a reflected version, arriving a few milliseconds later (and from a slightly different direction). An inverse filter could cancel this at the listening position (with a kind of inverted pre-reflection, say). But there are a couple problems: the pre-reflection itself gets reflected (so the cancellation filter needs to extend basically forever); and second, the cancellation will only work properly at the exact same position as the measurement. A few inches to the side, and you'll get an even worse situation than before. This effect is quite easy to hear in the filters that DRC creates. If you use a gentle filter (soft, erb, minimal, etc), the HF correction is extremely gentle, and over a really short time-window. But the stronger filters (normal, strong, etc) give more correction at HF: and the trade-off is that they're increasingly sensitive to the listening position. A different example: what if you put a big bell next to the loudspeaker? The bell is very resonant, and so very difficult to "invert" for. Now, rooms themselves are like bells - for frequencies below 120Hz or so, anyway - because they resonate as standing waves build up. And is the bell resonated at (say) 200Hz: it would make a huge spike in the measured 200Hz room response. But would you really want to just cut 200Hz out of the signal, to end up with a flat room response? Probably not - the bell sound would still be there, but the music would begin to sound plain weird. darrenyeats;228997 Wrote: > Now here is part of the theory behind omnipole and dipole speakers. They > spread sound in all directions, which aids in creating a sound field > which the brain is used to. After all, normal real sounds spread out in > all directions. With natural sounds their reflections - albeit arriving > later and attenuated - have a similar quality to the original. In > contrast box speakers, which have a directional output, naturally cause > reflections to have a different quality to the direct sound. (Stand > behind your box speaker. Even *before* the sound bounces off the back > wall it has a different quality from the sound in front!) Exactly right. I think the difference between the "direct sound" and the "room sound" is a big part of what makes recorded music less than "live"-sounding. The general wisdom seems to be (hmmmph... can't think where to find a reference for this...) that "room sound" is typically too bright. I used a pair of Gallo Ref3 speakers for a long while, which are nearly omnidirectional (and rather excellent). But they didn't have very much "heft" - not I think due to lack of bass, although perhaps a bit of midrange suckout due to floor bounce - but perhaps because they threw a lot of HF sound into the room. Partly as a way to try understand this stuff, I'm now listening to horn speakers - very directional indeed (at least in the 1k-5k range). I like them a lot, and the amount of "room sound" altogether is at least 6dB down from the Gallos. Listening way off the beam, the reflected sound is still quite lifelike. Anyhow, that room/direct balance was one reason I built the 'flatness' (http://inguzaudio.com/blog/2007/09/17/flatness/) control and the "_direct" filter set. I don't think there's a single "right answer". Experimentation is good :-) -- inguz ------------------------------------------------------------------------ inguz's Profile: http://forums.slimdevices.com/member.php?userid=1139 View this thread: http://forums.slimdevices.com/showthread.php?t=35615 _______________________________________________ audiophiles mailing list audiophiles@lists.slimdevices.com http://lists.slimdevices.com/lists/listinfo/audiophiles