pinnahertz
Headphoneus Supremus
- Joined
- Mar 11, 2016
- Posts
- 2,072
- Likes
- 739
1. Maybe the all rooms I've measured are just freaks of nature but I've never seen or even heard of room which does not have very significant variation in FR response throughout the audible spectrum, with high res, non-smoothed RTA measurements from about 1kHz upwards typically resembling a seismograph during an earthquake! Freqs at/below the Shroeder commonly consume much/most of the treatment budget in a studio build but we certainly never just ignore treatment well above the schroeder.
2. Your statement ("room modes affect lower freqs only") appears to contradict 2a? 2a. Agreed to an extent and even then, certainly not a flat/random reverb!
3. I did not say a 40dB dip in response, I said "peaks and troughs". Looking at the worst peak and worst trough, a 30-40dB variation between these two points is not only entirely common in my experience but something I've witnessed pretty much without exception in every untreated room I've ever measured. At the standard RTA resolution of 1/3 octave, I can't remember off the top of my head ever seeing anything quite so drastic but using a 1/6 or 1/12 8ve non-smoothed, to see more precisely what's going on, it's entirely common.
3a. Again, total variation between worst peaks and worst troughs at higher than 1/3 8ve resolution. Generally, there is at least 10dB between those points, even in world class studios! Pro studios are, as you mentioned, quite a different kettle of fish though. One obvious difference is a large reflective surface right in front of the LP (the mixing desk) and of course we can't solve this by just covering the desk with a trap or two (or with the use of additional subs)! Also, absorption panels (traps) are ineffective below about 80-100Hz, to treat the almost inevitable LF problems in any relatively small room requires treatment other than simple absorption traps; tuned panels, Helmholtz resonators, etc., treatments which are generally well beyond all but a tiny minority of the most extreme consumers.
4. Thanks, I'll try to look out for that paper but I've not been a member of the AES for quite a few years. That Shroeder freq tallies well with my experience but I'd be very surprised if the paper really concludes what you appear to be implying. Maybe I'm mis-reading but the implication I've understood from you post is that room response above approximately 200Hz is largely inconsequential. This would fly in the face of standard pro studio design, where we always treat well beyond the Shroeder freq, typically not just (and sometimes not at all) with broad band absorption (traps) but also with various forms of diffusers, to randomise the interactions of higher room mode harmonics/reflections. Additionally, commercial studios are specifically constructed to avoid certain quite severe acoustic problems, problems which are often inherent (without treatment) to consumer listening environments. Flutter-echo being an obvious example.
5. I agree. I wasn't talking about what is the most audibly objectionable form of distortion though, just very significant measurable variations from flat, as opposed to the difficult to measure, let alone audible, distortion of say ADC jitter. Most consumers, even many/most audiophiles are blissfully unaware of their acoustic problems, even less so above their Shroeder freq but that doesn't mean the problems don't exist or aren't audible. Trained listeners, such as experienced pro audio engineers, would be aware of these problems (including those an octave or 3 above Shroeder!), and the consequences (when surgically EQ'ing for example). I also wasn't speaking about extreme circumstances, pushing a system/speakers beyond their capabilities (towards Xmax for example), although I make no judgement about how common those extreme circumstances may be amongst consumers.
6. Unlike you, I have no experience of measuring turntables. I've never owned one myself and have only experienced other peoples'. In those cases, they were expensive well maintained units and while even the best turntables are obviously still susceptible to the deficiencies of vinyl (HF inaccuracies, etc.), I didn't realise the FR variation was generally as bad as you indicate.
If we consider that by the time we're 1m or so away from our speakers, what's hitting our ears is roughly a 50:50 balance of reflections vs direct sound, then acoustics is obviously a very major player and by two meters or more, easily the dominant player, at least as far as FR is concerned, if not conscious awareness. Certainly I've measured and heard $1k speakers in a well treated room which perform significantly better than speakers at pretty much any price in an average consumer's untreated room. I'm not really disagreeing with the principles you've mentioned, mainly just with degree and from my personal experience I would switch the positions between your #1 and #3. It's entirely possible/likely I was incorrect with my "degree"as far as turntables are concerned and of course we can come up with extreme examples which don't conform to my assertion that acoustics is the dominate factor of FR but I do stick to that assertion as generally true and that acoustics, speaker and turntable distortion is likely to add up to well over 1000x the distortion of a competent ADC.
1. No, the rooms you measured are likely not freaks of nature. However, your measurement technique is at fault. High res non-smoothed RTA (I'll bet also measured at a single point) is not representative of real room response. I'm not even sure how you'd respond to your measurements. Yeah, it'll look like a seismograph. If you must use an RTA (and I still do a lot of the time), you need at least use a little smoothing, limit resolution to 1/24th octave or 1/12th octave, then most importantly, average many measurements over an area that includes more measurements clustered in critical listening positions, but includes several at the edges of each of the LPs too. The technique is not new, it's the old "spacial/temporal average" thing. The point is you're seeing a lot of inaudible detail that is specular, confined to a single mic location. People don't hear that way, it's an anomaly of measurement only.
2. What I meant was that room modes affect measured frequency response at lower frequencies only. I think you did get it, and there's no contradiction.
3. My first comment would be the use of the RTA technique again. Using high res unsmoothed...you aren't actually seeing precisely what's going on, you're seeing excessive inaudible detail that nobody hears, because nobody's ears are the equivalent of a single omnidirectional mic on a stand. We have two ears, we move around. We always hear an approximation of a spacial average.
3a. Yeah, the console reflection. I know the problem well. Can't solve it with treatment at all, has to be a positional fix. But the effects of the reflection coming off a console surface wouldn't be into the sub range much at all, both because of wavelength and dimensions of the reflecting surface, and wavelenght vs time delay. Those properties place the effects of the console reflection in the mid-band and up to frequencies where the console becomes a diffuser, so it has nothing to do with subs anyway. You also wouldn't address the issue with traps of any sort, it's just not in that frequency range. The only means of addressing it is physical positioning of the console and speakers. Ideally the console goes away and becomes virtual, which is now possible, though still a bit impractical. Fortunately (or not, depending on your perspective) the issue doesn't exist in the home. It is possible in a fresh design to put the console reflection at a point where it lands away from the ears, it's just not often completely practical.
4. PM me if you can't find the paper somehow. It's based on over 500 rooms, multi-point time-domain based analysis.
5. I think we agree on this. I'd be very cautious about applying surgical-precision EQ above Schroeder without doing the multi-point average, though. Chasing tails, problems that aren't really there in more than one specular measurement. Consider a wavelength above 1kHz, and how little positional delta is required to move out of a deep null.
"If we consider that by the time we're 1m or so away from our speakers, what's hitting our ears is roughly a 50:50 balance of reflections vs direct sound, then acoustics is obviously a very major player and by two meters or more, easily the dominant player, at least as far as FR is concerned, if not conscious awareness. "
It would take a live, tiny room and speakers with very wide dispersion to get that 50/50 ratio at 1m. Above the transition reflections don't have the same effect on audible response as they would below transition because there are so many, and wavelengths are so small. Again, to see what's really audibly going on you have to get away from high res single position measurements, do a bit of smoothing and average multiple positions with a position delta of inches to feet. See "Sound Reproduction....", F. Toole, ch 4.3.1, "One Room, Two Sound Fields—The Transition Frequency", particularly Fig. 4.10 and discussion.
We may tend to EQ a live room by reducing HF energy, because a live room tends to sound bright. But that's not real frequency response, thats integration of multiple arrivals over time. The EQ solution doesn't really work.
In small rooms at home or in control rooms, the reverberant field is pretty easy to control. In fact, the cited paper in my previous post shows analysis of RT in hundreds of "untreated" rooms with a surprisingly short RT. When we consider what the "typical" room actually is, we can see that the dominant quality determining factor is always the speaker first above the transition frequency, room first below it.