That might well be but I went over the process several times and can't find any errors.
We're looking at an in-room response taken at the main listening position. Speaker is a Geddes Nathan 10.
I agree but at one point we have to get past educated guessing and find practical answers. That's what this thread is about.
I've been talking to Uli about this and Acourate indeed applies psychoacoustic filtering first followed by frequency dependent windowing.
Ears and microphones respond to continuously varying air pressure.
How human discriminates frequency is unimportant.
Magnitude and Q of filter that human can discriminate when applied to broadband signal is only important in terms of number of filters required to get useful result.
In theory a speaker's frequency response may be equalized to perfectly flat with a sufficient number of PEQ filters, regardless of the resolution level of the system used to measure the system's frequency response. Perfectly flat response equates to IR that becomes Dirac pulse. Such a construction of filters sums to be the inverse transfer function of the system response.
How human discriminates frequency is unimportant.
Magnitude and Q of filter that human can discriminate when applied to broadband signal is only important in terms of number of filters required to get useful result.
In theory a speaker's frequency response may be equalized to perfectly flat with a sufficient number of PEQ filters, regardless of the resolution level of the system used to measure the system's frequency response. Perfectly flat response equates to IR that becomes Dirac pulse. Such a construction of filters sums to be the inverse transfer function of the system response.
It's practically impossible to build a spatially robust filter that results in a perfect Dirac.
Our hearing doesn't work like a FFT. It's sensitive to how sound changes in time.
Member
Joined 2003
He doesn't disclose that information. The result looks like this:
red = magnitude response with large window
green = Acourate "Psychoacoustic"
Attachments
You keep saying how the ear/brain perceives doesn't matter but that still doesn't make it true.
I've already pointed out why you're wrong - for one, the ear does not sum the different reflections arriving from different directions equally like an omnidirectional microphone due to our HRTF. (Not to mention our perception is based on a melding of two different measurements spaced a head width apart)
Any correction based on an omni-directional microphone measurement which includes all reflections indiscriminately (no windowing) cannot correct the perceived response at high frequencies as it cannot predict the directional effects of the reflections on the summed response.
This is before you even start to consider temporal effects like the way the frequency response of the first arrival dominates our perception at high frequencies independently of the delayed reflections.
Beating the same drum of denial won't get you any further.
Last edited:
Even so, it still doesn't look right, unless the speakers have a very droopy axial high frequency response and a very flat DI. What does the windowed axial response of the speaker look like at 1-1.5 metres ?
I've taken the same type of measurements before at the listening position with variable window times on my speakers and although the high frequency response measured with a short window far away is not nearly as accurate as a 1-1.5 metre measurement (window time to first reflection is too short for good frequency resolution) the trend was clear.
At the listening position with a very long window the high frequency response slope is pretty close to the room/speaker power response, (eg sloping down about 5-10dB from 2Khz to 15Khz, can't remember exact figures) while the short window response is fairly close to the free field axial response of the speaker, especially above about 5Khz - eg more or less flat +/- 1dB or so.
My concern is there is NO apparent change in the slope of the high frequency response with changes in window time it just looks like progressively more smoothing was applied. Do you have some more conventional speakers you could measure as a comparison ?
Maybe speaker dave will pipe up with the name of the papers - the hard work of figuring out the correlation has (probably ?) already been done, although from what I remember it was an AES paper so I didn't have access to actually read it...
Last edited:
You keep saying how the ear/brain perceives doesn't matter but that still doesn't make it true.
I've already pointed out why you're wrong - for one, the ear does not sum the different reflections arriving from different directions equally like an omnidirectional microphone due to our HRTF. (Not to mention our perception is based on a melding of two different measurements spaced a head width apart)
Any correction based on an omni-directional microphone measurement which includes all reflections indiscriminately (no windowing) cannot correct the perceived response at high frequencies as it cannot predict the directional effects of the reflections on the summed response.
This is before you even start to consider temporal effects like the way the frequency response of the first arrival dominates our perception at high frequencies independently of the delayed reflections.
Beating the same drum of denial won't get you any further.
But how are you able to record anything then (one of the steps before playback is possible)? We don't use our ears in the recording process, usually microphones do that part. So microphones should be able to record the playback as well, apart from the differences of all speaker types that are in use.
Very different situations.
For starters on most recordings there is no original acoustic space to capture - you have multiple microphones all close mic'ed with each microphone picking up only one maybe two instruments and very little room contribution. Room reverb gets added in post processing. So there is nothing "real" for you to compare the reproduced sound to. It sounds however the recording engineer decided it should.
In a more minimalist recording technique where a real venue is being captured standard stereo mic'ing is already throwing a lot of directional information away. Record the venue with a standard stereo mic or displaced microphones then record the same thing using binaural technique with a dummy head.
Do they both capture the same thing ? no!
The typical stereo mic set up has made a very different recording than a binaural recording set up, so why should we consider the standard mic's to be "correct" ? We have just grown used to the lack of spatial fidelity in stereo recordings - in a multi-mic recording some of the spatial fidelity can be "recreated" in the post production process using various tricks to place the different sounds so that what we hear has more spatial "realism" than a raw recording would have, but the point is the standard microphones did not capture the room as we would have heard it.This is all very different to trying to measure/correct a room though. Before you can correct it to sound right you have to model how we perceive sound. An omni-directional microphone does not model the human hearing system. Not even close. Then you need to answer the question, should the ROOM be corrected at high frequencies, or should only the SPEAKER be corrected at high frequencies ? That will then dictate the measurement technique used to derive correction filters.
Last edited:
No but at least in Acourate the programmer tried to mimic important psychoacoustic phenomena of our hearing in the capturing process. One thing that is clear in the above pictures (I take it this is the first step, measuring the data) that he disregards the narrow dips from the microphone pickup for further processing. Combined with the sliding window (the shorter the waves, the shorter the window captured) he is at least trying to get closer.
The omnimic does not have to model the human hearing, that was the task of the engineer shaping the sound after the recording session. We are not trying to recreate the original event in the case you mention above but the shaped event the sound engineer created in that last step.
When he played it back on his speakers/headphone etc to make that end product.
Seems to me you should be able to get close to that recording with the omnimic, if besides SPL the timing is included as a factor in the measurement. It is up to the end user in this case to shape the sound for his room/speakers. That would be the curve you draw to process the sound to your liking in the case of Acourate.
Shure, I'm not saying a microphone hears what we hear but what is the alternative? With the microphone we can analyze at least a big part of that signal to what was in it to begin with. The rest will be up to us.
What do you suggest we must do to get closer?
There is no telling what speakers and other equipment was used to shape the sound that's on our recordings but that won't make me disregard the music I own.
The omnimic does not have to model the human hearing, that was the task of the engineer shaping the sound after the recording session. We are not trying to recreate the original event in the case you mention above but the shaped event the sound engineer created in that last step.
When he played it back on his speakers/headphone etc to make that end product.
Seems to me you should be able to get close to that recording with the omnimic, if besides SPL the timing is included as a factor in the measurement. It is up to the end user in this case to shape the sound for his room/speakers. That would be the curve you draw to process the sound to your liking in the case of Acourate.
Shure, I'm not saying a microphone hears what we hear but what is the alternative? With the microphone we can analyze at least a big part of that signal to what was in it to begin with. The rest will be up to us.
What do you suggest we must do to get closer?
There is no telling what speakers and other equipment was used to shape the sound that's on our recordings but that won't make me disregard the music I own.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.