Member
Joined 2003
Think I missed something. This also supports Pano's suggestion a while back that there is an awful lot of information on conventional media that we somehow do not audibly reproduce.
Yes, probably. In this case it seems that it was getting reproduced, just swamped by massive reflections.
I believe that diffusing those reflections so that they are less coherent can really help.
I believe that diffusing those reflections so that they are less coherent can really help.
Paul,
I haven't had time to reply to this thread last couple days. I also don't know of online resources. I learned this data the old fashioned way, through textbooks and professional Journals (JASJ, JASA etc). I used the concepts here:
Method and apparatus for inhibiting echo in a channel of a communication system - Patent 6178162
Dave Dal Farra
@DDF
I like that summary too.
But subjective enhancement of 'depth' is not explicitly
mentioned and 'spaciousness' is not the same as 'depth' to
my mind.
What is spaciousness without depth ?
Which (listening) room properties may contribute to the sense of depth ?
@Pano
You wrote in an earlier post, that you suspect a lot of (spatial) information even
on common recordings is not really reproduced ...
I like to pick up that strand and answer the questions from above from
my personal point of view, without claiming scientific correctness:
What is spaciousness without depth ?
Nothing. If the image is flat, there are still two "speaker clouds" especially
with mediocre recordings which lack consistent cues for reverberation from
the recording venue.
Which listening (living) room properties may contribute to the sense of depth ?
None. The small listening room will always tend to flatten the image. The only way
is keeping reflections as diffuse as possible or avoid (especially early) reflections
as good as possible.
I believe the sense of depth has much to do with the (binaurally) decorrelated
field, which usually can only evolve in a larger hall having appropriate acoustics.
That "large room diffusivity" can neither be faked nor improved, if it has been
captured somehow on a recording.
Without indtroducing diffusivity, even rather directional speakers like horns (CD Waveguides),
dipoles, line arrays produce cues indicating the proximity of the listening room's walls.
Also dependent on the type of speaker you will have to choose between
- removing the listening room's walls
- making the walls reflect in a diffuse manner
- or use heavy broadband absorption on the front wall, which may be useful if the
speaker has changing directivity from lower midrange to highs ...
The pistonic acting speaker can be made having a smooth dispersion with reasonably flat
on axis response and consistent off axis response. It can also have reasonably flat
group delay and thus be able to reproduce the direct sound on a recording with low
coloration and sufficient precision in the time domain.
But it will ineviteably fail to transport the 'depth' information present on many
recordings, because it will produce correlated reflections in the listening room,
which are evaluated as "non conforming" to the diffusive spatial information on a well
made recording.
Currently only panel form bending wave speakers are able to produce decorrelated sound
for off axis angles. It is not important for the listener, whether a wall is hit
by a diffuse "wavefront shape" (a contradiction it seems) and now reflects that
diffuse wave form in a non diffuse manner (it will still be diffuse after the reflection)
or if a stronger contoured wavefront hits a wall which has been prepared to reflect
in a diffusive manner.
The listener has to be "protected" from binaurally correlated reflections bouncing
from the listening room's walls: Then you get that depth you have described
- which can extend several meters behind the speakers plane - and make you feel you
could walk right into the scene.
That kind of depth i am talking about does not conflict with definite phantom source
localization, "crisp" transients coming from instruments which produce those
and it does not even need that listening room's "substitutional helper spaciousness",
which is only desirable as long as you try to reestablish what is lost from the recording's
spatial and depth information.
I can imagine very well how your "sans front wall" experience with dipole speakers was like.
And that is what i want too, nothing else.
Cheers
Last edited:
@DDF
I like that summary too.
But subjective enhancement of 'depth' is not explicitly
mentioned and 'spaciousness' is not the same as 'depth' to
my mind.
What is spaciousness without depth ?
Which (listening) room properties may contribute to the sense of depth ?
What follows below is Humdingers Opinion (I'm still learning how to do this - sorry):
Spaciousness to me is about the X and Y axis of sound, while Depth is more about the Z axis. All three axis obviously work together in the real world. In a reproduction, some of the spacial and depth information may be "freeze dried". Huh?!
In the world of 2 speaker reproduction, we pretty much only get one axis through (the X), and then depend on the listening room reflections to regenerate the Y and Z axis. It fascinates me how well the mind can reconstitute the whole Auditory experience in apparent 3-D.
From my experience, significant amounts of typical size listening room reflections actually enhance the sense of depth and spaciousness in a recording. Theory seems to suggest that an anechoic chamber should be better. Why isn't it? It seems that when you add some listening room reflections, the embedded reverbs and timing cues in general, embedded in the recording, come to life more. It's like you've given them their other two axis back, using the listening room acoustics. Huh?!
My Holographic Generator circuit does a partial cleanup of inter-aural crosstalk on the X axis (unless side walls are too close in), thereby spreading the perception of images out beyond the boundaries of the 2 speakers and somewhat cleaning up image individuation. The fact that my speakers are of open baffle design means they do less emission into the room on that same X axis (side to side), thereby making the cleanup job for the Hologram circuit less encumbered (less side wall reflections), while generating "reverb" primarily on the Z axis. The Hologram reconstitutes spaciousness left to right, and the Open baffle room reflections reconstitute the sense of depth and perhaps to some extent height.
Last edited:
Many good questions but time is tight. Lateral detection is enhanced because of the symmetry of the head and that we have two ears on the sides. Using our two ears, we can detect differences between them, but these differences are most enhanced by sound sources located to the side (ie side wall reflection). For sound sources in front, up or down, there is no difference presented to left or right ear.
We use the outer pinna to detect tonal differences from up or down as outer pinna is not vertically symmetrical.
As for the ability to detect reflections, and the effect of the reflections delay on detectability, there are many studies. I'll quote a quick result, from Begault "Audible and Inaudible Early Reflections: Thresholds for Auralization System Design: AES 1996 conference in Copenhagen.
Simulating lateral reflection at 30 degrees (~ 5 ms delayed for small room, 13 ms for large room) and the floor (3ms delay), first reflection threshold for detection was -19.5 dB for side wall, -14.5dB for floor. This number isn't gospel as there was significant variance between listeners.
I have a stack of conference papers on this topic, but the kids are calling out for a Dairy Queen run so ...
Dave Dal Farra
Other Dave here...
AES papers by Salmi were pretty good on this, but the most sophisticated one seems to be the paper by Kates:
AES E-Library: A Perceptual Criterion for Loudspeaker Evaluation
David S.
I read the question as ear integration time, not forward masking, but the Kates paper is good. Its limitations are that it uses a monaural model and the experiment producing the data (from Atal etc) isn't defined (headsets? speaker? if speaker direction? I'm guessing this is for headsets, in mono) From my notes on Kates:
- highest sensitivity (frequency response changes) to reflections is with 2 to 5 ms delay
- Threshold of perception of colouration for a single repetition:
no delay: -12dB
2to 5 ms: -18dB
10 ms : -16dB
16 ms: -15dB
20 ms: -12dB
30 ms: - 8dB
40 ms: - 6dB
- Kates supposes that lateral reflections have less effect but this was shown to be untrue in later studies
I didn't post about this study since the model presumed assumes lateral audibility is less than frontal, which seems counter to the more modern data I've seen. Its true for frequency response change, but not for absolute detection (which includes imaging/spaciousness).
Dave Dal Farra
Last edited:
Here is some more insight and good summary of the localization concepts... not so heavy reading as the AES stuff for the rest of us mere mortals. 😉
from http://www.silcom.com/~aludwig/EARS.htm#Source_location
from http://www.silcom.com/~aludwig/EARS.htm#Source_location
In my understanding, this puts the whole notion that our brains can't interpret differences of less than 10ms into question....doesn't it? (The distance between our ears is only about 0.5ms)
yes - timing has to make the difference and in some cases perhaps also phase - spectral results of combing patterns
and thanks for interesting link to Dr Usher
I would rather say that the properties of a typical audiophile overarranged and overdamped room will always tend to flatten the image
but this is not not neccesarily truth in a case of a typical living room
this is also my experience, and an interesting theory
best,
graaf
What I like about the Kates paper (and Salmi's) is that he comes up with a concrete perceptual hearing model. He uses knowledge on the hearing mechanism, including neural factors to come up with a measuring scheme. Like most who have explored that, it is a combination of frequency dependent time windowing and critical bandwidth filtering plus other elements. The key here, which gets back to the essence of this thread, is that if you have instrumentation that mimics hearing you can measure a speaker in a room and say: "thats what it sounds like". If it measures flat it sounds flat, if it doesn't then it won't.
In this recent discussion lets remember that we are talking about (sometimes confusing) two different things. One is the perception of frequency balance in a live room. The second is our perception of the spatial environment (depth, envelopment, etc.). Most of these studies use reflections of various delays, locations, balances, and plot out threshold of changes of: level, frequency balance (timbre change), lateral position, image source width, etc. The shifts of frequency balance point the way to measure "what is flat". Of these my understanding is that frequency balance shift is more commonly an attribute of frontal reflections (on the plane between the ears) and less so for lateral reflections. This would justify a mono model.
The Kates numbers suggest strong detectability in the 100 to 250Hz range which again points to the floor bounce, or perhaps the front wall bounce when speakers are close to the boundary. Both time windowing and critical bandwidth tend to ignore HF room reflections, tend to admit all LF room effect and somewhat limit mid frequency effects to the first couple of reflections. This is consistent with everything I have read on the subject.
David S.
I have a question about how "threads" work. I post and then get a notification about further posts, but only for a few times, then all notices stop. I eventually come back and find that I missed a lot of posts. Is that because the thread moves so fast? Is there some way to always get notified of any posts to a particular thread?
Moulton in one of his AES papers reported that lateral reflections also have a significant impact on timbre if the reflection had a spectrum notably different than the direct sound. If the lateral reflection and the direct sound spectrum were equivalent, the comb filtering was inaudible. He also found that the precedent effect was ignored in the higher frequencies, supporting your observation.
So, the speaker's DI and placement also have to be considered. If the response that illuminates the first side wall reflection deviates from the direct sound, you'll hear it as a change in timbre, according to Moulton's tests.
Dave Dal Farra
Mentions it in the notification. Once you get a notification, you'll get none further unless you repost.
Thanks for that very interesting link.
I had done all this same work when looking at how to improve the LF sound field and found that decorrelation of multiple subs could almost completly elliminate any modal effects. My only comment on the paper is that did not seem to realize that electronic reverberation is basically a decorrelation filter (back in 95 these may not have been too common) - depending on the settings. Hence using a reverb on just the LF signals in a room (multiple subs assumed) would have the effect of decorrelating these signls and minimizing the modes. I simulated this once, but I have never tried it in a real situation. I'd like to sometime, but I spend too much time writting in threads.
Actually you don't have to post a further message - viewing the thread in your browser (as long as the browser is logged in to your diyaudio account) is enough to prime it for another notification email upon the next post to the thread. If the thread was already open in the browser doing a page refresh/reload is all that's needed.
However any posts that people make after a notification email is sent to you but before the next time you view the thread don't get email notifications, this is normal for this bulletin board software - it's to prevent you from being flooded with emails when you're not actively viewing the threads in your browser.
Another option is daily "digests" which always contain all postings for the day, all in one email, regardless of whether you've already read them or not, however you then lose the instant notification of new posts.
Last edited:
This statement implies that if the response is the same that it is not heard as a change in timbre. I don't think that is correct.
The first wall reflection is a critical one and if it exists to any appreciable degree then it is audible, usually as a timbre change. This is especially true for the first left reflection from the left speaker to the left ear (and the same for the right ear). In this case the signal actually arriving at the ear cannal has had its timbre changed. The reflection off the right wall to the right ear from the left speaker is not as highly correlated and the brain has some change of sorting out that one is a direct signal and one is a reflection. In the prior case this is not possible, these nearly identical signals are mixed and interfere external to the ear so no brain processing can possibly take out this effect..
So if I read the post in Outlook but do not actually go to the thread, it will stop sending me notifications? I read a lot of these posts on my phone and its not very easy to respond (nto at my age on that small thing), but its easy enough to read, and then if I don't have any comments I just delete them. Then, if what you are saying is true, I won't get any more notifications.
Thats not the way it should be for a fast moving thread like this.
- Status
- Not open for further replies.