Originally Posted by Floyd Toole
This is a large topic. Look at Figure 2 in my 2015 JAES paper "The measurement and calibration of sound reproducing systems" (it s a free download from aes.org, click on publications, click on open access, type Toole and download). It shows the DIs of a typical cinema speaker, the M2 and a Revel cone/dome system. The constant-directivity horns are instantly recognizable, as is the result of different dispersion design goals - the M2 is designed as a moderately wide dispersion monitor for small rooms. The cinema speaker is more directional, designed to deliver sound to the audience without energizing walls and ceiling. The cone/dome system inevitably has lower DI (wider dispersion) and a progressively rising DI. These are safely generalizable observations. In a normally reflective room there will be audible differences attributable to DI.
Figure 18.22 in my book shows a comparison of an NS-10M and a JBL 4301, both 7-8-inch two way systems. The NS-10M was designed to have flat sound power output (a mistake), so the on axis response is lumpy, the JBL was designed for flat on axis response and so the sound power is lumpy. The DIs are almost identical, and not very flattering compared to multi-way or horn systems. An 8-inch two-way speaker is inevitably going to be room sensitive, especially if the tweeter has no horn/waveguide, which is common. It will do better in a near-field situation where the room matters less.
Thanks for the reply! Sorry for the delay in reply. I realize this conversation is a bit stale.
For those who are following this discussion, here is a direct link to the AES paper described above:
That paper has a lot of interesting information. I actually read it before along with the related SMPTE report
. I am actually very interested in the subject of achieving consistent playback between rooms, and I have some ideas of my own about how to approach that problem. That deserves its own discussion later, but for now I'll just say for now that I believe trying to fit so-called "steady state frequency response measurements" to some kind of target curve is fundamentally unproductive.
But I want to return to my questions about ultra-high frequency response, because I don't think you really answered them. The basis for my questions is the fact that most speakers have high DI slope vs. frequency above 8 kHz or so to the point that the pattern often becomes much narrower than the listening window. The obvious problem here is that you don't have flat direct sound across the listening window anymore. You are forced to make a compromise when dealing with these ultra high frequencies (UHF).
So when you are forced to compromise, is it better to optimize for flat on-axis (OA) or flat listening window average (LWA) response? As I said, the Salon 2 appears to prioritize OA, but the M2 appears to prioritize LWA. In both cases, "flat direct sound" is the guiding principle, but very different compromises appear to have been made for the frequencies in which that principle could not be completely honored. I could be wrong, but my guess is that the M2 will sound hotter at the top than the Salon 2 will when listened to on-axis.
Originally Posted by Floyd Toole
The small differences at HF are of course audible - depending on the choice of recording. Many recordings are a bit "hot" because of where the mikes are placed - e.g. a few inches from a vocalist's mouth, close to strings, etc. Some people like it, and accept it as part of studio-created music. Mastering engineers might or might not tame it. Years ago Sean Olive and I noticed that some recordings had colored HF and decided to investigate studio mikes. We found that some widely used large diaphragm mikes had significant resonances at HF - in fact the frequency responses would not have qualified them for use as tweeters in modest speakers. Now there are numerous superb small diaphragm mikes, but the old warhorses still are still used - a trade-off of virtues and deficiencies. It is all in the art. Just monitor through neutral speakers so these characteristics can be heard.
Classical recordings are often made with elevated mikes which pick up more HF from the violins than is heard by the audience. Some classical recording engineers have favored loudspeakers with a slight depression in the frequency response that compensates for this - even arguing that such loudspeakers are better for classical music. This regrettable practice means that those of us with neutral speakers might complain about strident highs - their monitor speakers are part of the EQ.
When art and technology are so tightly bound confusion exists. Unfortunately many artists mistrust science.
I think the inconsistency of the monitoring environment is the only thing worth talking about here. And it's a *big* issue with UHF content in particular. From what I have heard, most monitoring systems use UHF roll-off of some kind, and most existing content was mastered with the expectation that it'll be played back on systems with UHF roll-off. I know that many monitors come with UHF roll-off baked in.
Obviously, the fact that UHF roll-offs are in widespread use but are not standardized in any way means that UHF monitoring conditions are all over the map in the real world. On the plus side, depending on the shape, a UHF roll-off may not change the timbre as much as one would think, or so my ears tell me. I have a hypothesis about that, and I use roll-offs with a particular curve shape based on that hypothesis to try to minimize timbrel changes while allowing me to better accommodate a variety of content.
For my M2 evaluation, this adjustable roll-off was not available to me, so I purposely chose some music selections that, to my ears, sound natural when run with flat UHF on my own system. This is the best I could do to deal with the Circle of Confusion. I know it was still not enough to make a reliable judgment. The M2s may sound fine, and this minor flaw may only exist in my imagination.
Actually, I thought the M2 sounded at least as good as any other speaker I've heard. The only thing comparable are my own speakers. The lack of resonance and harshness on the top end made whatever excess was there (or wasn't) very inoffensive. As I said, I can see a lot of people really enjoying its top end sound, even if it isn't completely neutral.
The only concern I would have is if the extra bit at the very top was excessive enough to substantially mask lower level content. I have observed this phenomenon with UHF (especially near 20kHz and even above), and it is very insidious. That kind of sound can "eat the subs", masking away almost all perception of the bass. I doubt this would be an issue with the M2, but I couldn't judge this with certainty given the circumstances. (John's room EQ system was out of calibration, and we had to rely on no room EQ.)