Hello,
In theory that would be great because one could assume that would generate more comparative measurement results of different systems. But nobody seems to optimise the time response. What I've seen to be done they use white noise and long averaging to optimisation.
Nobody interested to check the time response?!? Amazing.
Well, to your post to have any relevance you should change "we" to "I" where the "I" is meaning you. Transients below 100Hz are very easy to hear.
Have you ever done the Linkwitz test? What you are listening to is the modulation of the envelope. That is the transient.
http://www.linkwitzlab.com/images/graphics/mlt-bst1.gif
About ten years ago I was thinking to make similar system like DBA but sideways for stereo. But the WAF of such things is too bad.
- Elias
gedlee said:
In theory that would be great because one could assume that would generate more comparative measurement results of different systems. But nobody seems to optimise the time response. What I've seen to be done they use white noise and long averaging to optimisation.
Nobody interested to check the time response?!? Amazing.
gedlee said:
Well, to your post to have any relevance you should change "we" to "I" where the "I" is meaning you. Transients below 100Hz are very easy to hear.
Have you ever done the Linkwitz test? What you are listening to is the modulation of the envelope. That is the transient.
http://www.linkwitzlab.com/images/graphics/mlt-bst1.gif
KSTR said:
About ten years ago I was thinking to make similar system like DBA but sideways for stereo. But the WAF of such things is too bad.
- Elias
Elias said:
I have no idea what the attached link is supposed to show, but you would have to go a long way to convince me that we can hear a transient for a sound with a period of 10 ms or more. And hearing the envelope is not the same thing as a transient. From all that I know what we sense at LF is a steady state sound. There were studies of "modal decay" in the AES which showed that people did not hear these decays. They heard the integrated sound level, i.e. a more steady state sound.
Same freq. as im my previous post, this time with a short shaped burst. Doesn't look like a single reflection from a simple floor/ceiling bounce. More like a repeated reflection that will become a mode when the exitation is long enough. The shorter pulse has wider spectrum, though, which mus be taken into account.
The Linkwitz graphs is just exactly what Elias and I are concerned with. My understanding (and perception) also is that we hear the envelope pretty well. Also we know that in the bass region the hair cells fire only during the positive (relative) pressure halves of a waveform, which explains a lot of otherwise unexplainable phenomena (like sub-harmonics, ear "IMD", and audibilty of absolute poarity with transient asymmetric waveforms -- the latter I recently backed up with a clean ABX blindtest using sort of an "electric bass drum" as test signal).
- Klaus
The Linkwitz graphs is just exactly what Elias and I are concerned with. My understanding (and perception) also is that we hear the envelope pretty well. Also we know that in the bass region the hair cells fire only during the positive (relative) pressure halves of a waveform, which explains a lot of otherwise unexplainable phenomena (like sub-harmonics, ear "IMD", and audibilty of absolute poarity with transient asymmetric waveforms -- the latter I recently backed up with a clean ABX blindtest using sort of an "electric bass drum" as test signal).
- Klaus
Attachments
gedlee said:
The top trace is the signal and the bottom trace is what the mic recorded. The first one is at a frequency where there aren't any significant room effects and the recording is pretty clean. The others are at frequencies where there are significant room peaks or notches and they don't look so pretty.
Try it for yourself. Download and play http://www.linkwitzlab.com/images/sound/100mbst4.wav . It's "a signal consisting of four bursts with cosine envelope over 10 cycles of a 100 Hz sine-wave", repeated 3 times. If you can distinctly hear the 4 bursts, your system is doing okay at 100 Hz. If it sounds like one long burst, it isn't doing so well.
Ok, I tried it. I could hear the modulation on all the systems that I have, including my PC speakers. SO what? The bass in the PC speakers is awful and the bass in my HT is superb and this test came out the same in both situations. I simply don't get the relationship between this test and anything relevent about the subjective perception of bass.
And lets go one step further. In an auditorium the reverberation time could be seconds long and this signal will get almost completely lost in the reverberation. Does this mean that the bass in ALL large auditoriums is bad?
Sorry, but the real situation is just not that simple.
And lets go one step further. In an auditorium the reverberation time could be seconds long and this signal will get almost completely lost in the reverberation. Does this mean that the bass in ALL large auditoriums is bad?
Sorry, but the real situation is just not that simple.
If you only have the recorded sound and not any visual clues this could sound pretty bad -- even though due to the long reverberation time the effect is more like a rather slowly changing overall noise floor that is decorelated to a greater extent (compared to normal listening room sizes).gedlee said:
On *this* we'll most probably all agree, don't we?
- Klaus
The structure of my test signals show the effect even more pronounced than Linkwitz'. This is because of the longer/stronger exitation and the nature of the shrinking gaps between the individual bursts, plus the phase relationsships I mentioned.catapult said:
And (@Earl), all this types of signals are not too far away from real music signals unless one were very limited in genres.
- Klaus
I think the fact that we can all hear the bursts is significant. It shows that there's more to bass than the steady-state condition. The bursts are similar to certain musical sounds so it's not an unrealistic test.
The next step is to compare 'good' and 'bad' systems/rooms and actually look at the traces on a scope to compare the waveforms, testing over a number of different frequencies. It gives a better test than just 'can I hear it.' Computer speakers don't really count as a 'bad' system because you're listening in the nearfield where the sorts of room effects being tested are minimal.
The next step is to compare 'good' and 'bad' systems/rooms and actually look at the traces on a scope to compare the waveforms, testing over a number of different frequencies. It gives a better test than just 'can I hear it.' Computer speakers don't really count as a 'bad' system because you're listening in the nearfield where the sorts of room effects being tested are minimal.
catapult said:
This is a good point, I will do the test again.
I am not saying that this type of signal is meaningless at revealing some aspects of the problem, it's the interpretation of the results, as given here, that I have a problem with. I do almost nothing subjective, except listen to music. I am most interested in objective means for quantifying our perceptions. These signals may have some usefulness in that regard, but maybe not. Noe the less, the justification for their use in objective assesment is not there at the present, from what I can see.
I only partly agree. The problem is IMO that these kinds of narrowband signals have so many parameters that it is not very easy to find the best possible suited signal structure that could be assessed both manually and could be analized by some form of post processing giving results consistent with actual perception, along the lines (but not restricted to) the idea that Elias has presented. "Best possible" to be read as giving the most stable results under varying circumstances. I pretty much appreciate his work and see it as a starting point for further studies. After all the goal is to minimize or flatten out all modal effects as much as possible, with all means we have. Those test signals are also very nice to find the best places for absorbers etc. I personally used them even for recording situations to find mic and source positions that don't suffer from modal problems.
I found that with my signals it is interesting to listen what happens to the perception with or without any background noise. There is a big difference between complete silence (at night, using a cdplayer) and some broadband background noise like PC fan/hdd noise etc. With background noise the sensitivity to the modal phenomena is better as one can listen to how the noise gets masked in different ways depending to what is happening in the gaps between the bursts. In fact I have CD versions with extra noise added in for loactions/setups where it is too quiet.
- Klaus
I found that with my signals it is interesting to listen what happens to the perception with or without any background noise. There is a big difference between complete silence (at night, using a cdplayer) and some broadband background noise like PC fan/hdd noise etc. With background noise the sensitivity to the modal phenomena is better as one can listen to how the noise gets masked in different ways depending to what is happening in the gaps between the bursts. In fact I have CD versions with extra noise added in for loactions/setups where it is too quiet.
- Klaus
gedlee said:
Earl, the thing is none of your systems may have a problem at 100Hz so, of course, the test won't show any problems.
SL's graphs show pretty clearly what happens with a bad frequency in a bad speaker/room combo. I think those graphs are from his original AES paper on the testing method. Elias is just building on his work, plotting energy/time curves (time being expressed in periods) of all the frequencies on one graph.
The original testing-method paper is this one from 1980.
Siegfried Linkwitz, Shaped Tone-Burst Testing, JAES, Vol. 28, No. 4, April 1980
Abstract:
http://www.linkwitzlab.com/publications.htm#13 - Shaped Tone-Burst Testing
Siegfried Linkwitz, Shaped Tone-Burst Testing, JAES, Vol. 28, No. 4, April 1980
Abstract:
http://www.linkwitzlab.com/publications.htm#13 - Shaped Tone-Burst Testing
KSTR said:
I agree, what does the time domain have to do with that? Basically, if the LF response is as smooth as possible it will likely have the lowest decay rate as well (although that requires Minimum phase to be absolutely true and you can't prove that the LF stuff is MP). Discussing how any one type of source acts in time does nothing towards the goal of a smooth LF response.
gedlee said:
Actually, as we discussed before Earl, in a typical room the source to listener transfer function is typically not minimum phase. The result is that tests such as those presented here do not correlate to amplitude response at all. It is not even clear that the smoothest amplitude response would necessarily yield the "best" transient, what ever that is?
IME the very lowest order modes (say fundamental and second) are pretty close to MP. Then my goal is to setup the speakers and listening pos so as to have only boosts, not notches at those freqs and then apply reverse (MP) notch filters. The higher order stuff which is clearly not MP must be dealt with by room treatment and, again, optimized placement. Finally some gentle (broadband) overall EQ to get the amplitude response right.
- Klaus
- Klaus
john k... said:
Hi John
In general I agree, but the lower the frequency the closer it will be to MP. My comments here are stricktly limited to the very discrete modal region where the system is becoming close to MP. Much of the discussion is about HFs where everything changes, but my commenst are limited to the lowest frequencies.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.