Apart from the HRTF, in room response and a few other things talked about in this thread, there is another very important reason "Why Flat is Inaccurate."
Playback level. Change the playback level and the FR changes. Not the system - your ears! Remember the Equal Loudness Curves? Move that volume knob and the subjective FR has changed. That's a little of what I was alluding to in my previous post.
Why should that matter? Ever see a "Loudness" button on a preamp?
You have no way of knowing at what level the recording was mixed and mastered. There is no standard. (but there are some habits)
So you want to hear it the way it was intended? How are you going to do that? Even in an ideal case in which your speakers at your listening position match the FR of the mastering suite, what was the listening SPL there? If you don't have the exact same SPL, you won't hear it the way they did - the Equal Loudness curves get in the way.
At least in cinema sound there is a standard. Given a -18dBFS modulation level on the film you are supposed to achieve 85dB C weighted SPL throughout the main listening area. It's part of the alignment process in cinema sound and a SMPTE standard.
We don't have that in "musical" audio. You have no idea how loud it was in the mixing and mastering suites. You can only guess. But if you are not at that level, it will sound different - no doubt about it.
I like to set my reference level to be 85dB SPL at the listening position with pink noise in one speaker. I use a pink noise signal recorded at -14dB RMS as suggested by David Moulton. But you could (and I have) used the SMPTE -18dB. Who knows what was used in the mastering suite? You can only guess.
If we wanted a more accurate system, then the mastering SPL should be encoded into the recording as a meta-tag. Your playback system could read the tag and apply a loudness curve to bring the FR into line with what was heard during mastering. You would need to calibrate your system for this to be effective.
So you see, there are some very basic issues that insure that "Flat is Inaccurate." SPL is just another.
Playback level. Change the playback level and the FR changes. Not the system - your ears! Remember the Equal Loudness Curves? Move that volume knob and the subjective FR has changed. That's a little of what I was alluding to in my previous post.
Why should that matter? Ever see a "Loudness" button on a preamp?
You have no way of knowing at what level the recording was mixed and mastered. There is no standard. (but there are some habits)
So you want to hear it the way it was intended? How are you going to do that? Even in an ideal case in which your speakers at your listening position match the FR of the mastering suite, what was the listening SPL there? If you don't have the exact same SPL, you won't hear it the way they did - the Equal Loudness curves get in the way.
At least in cinema sound there is a standard. Given a -18dBFS modulation level on the film you are supposed to achieve 85dB C weighted SPL throughout the main listening area. It's part of the alignment process in cinema sound and a SMPTE standard.
We don't have that in "musical" audio. You have no idea how loud it was in the mixing and mastering suites. You can only guess. But if you are not at that level, it will sound different - no doubt about it.
I like to set my reference level to be 85dB SPL at the listening position with pink noise in one speaker. I use a pink noise signal recorded at -14dB RMS as suggested by David Moulton. But you could (and I have) used the SMPTE -18dB. Who knows what was used in the mastering suite? You can only guess.
If we wanted a more accurate system, then the mastering SPL should be encoded into the recording as a meta-tag. Your playback system could read the tag and apply a loudness curve to bring the FR into line with what was heard during mastering. You would need to calibrate your system for this to be effective.
So you see, there are some very basic issues that insure that "Flat is Inaccurate." SPL is just another.
thanks guys
Rob, good suggestion, I'll have to get myself a copy, it seems to cover lots as it is often reference in relation to room acoustics as well.
Until then, can you recall any 'rule of thumb' that covers the (average/usual) change/alteration of the FR from anechoic to room??
Pano, the volume aspect I too have found to be of extreme importance, if not in the theoretical aspect of 'hearing it as the engineer mastered it' then most definitely in the sense of it not sounding right at various levels, but often it just clicks at one level as opposed to another.
I, as a natural tendency, would tend to listen on the louder side than softer. Some recordings can handle being cranked (louder than I would suspect the engineer listened in the control room), yet most sound horrible at higher levels. That is leaving aside things like compression and distortion which become more obvious at higher levels, I'm more talking about the 'rightness' of the sound at a specific level.
I've only just started reading the mouton site, lot's of very good stuff in there. He certainly makes the point about having knowledge of the playback level that should be used as you say, but I'm not sure how realistic that is. He mentions that a good system should have ample dynamic range to handle the the peaks on a symphonic recording or indeed a heavy rock recording. In the case of say heavy metal, the average level would be pretty high, with room for the peaks as mentioned.
Yet he also mentions when mixing that (as they can do it for hours on end) the average level would probably be lower than it would be for lsitening, which tends to overide the notion of 'playing back at the same SPL as the mixing level', yet I feel the earlier thought of it gelling at the right volume definitely applies.
Probably missing something in what he says but hey, that is not unusual!
Rob, good suggestion, I'll have to get myself a copy, it seems to cover lots as it is often reference in relation to room acoustics as well.
Until then, can you recall any 'rule of thumb' that covers the (average/usual) change/alteration of the FR from anechoic to room??
Pano, the volume aspect I too have found to be of extreme importance, if not in the theoretical aspect of 'hearing it as the engineer mastered it' then most definitely in the sense of it not sounding right at various levels, but often it just clicks at one level as opposed to another.
I, as a natural tendency, would tend to listen on the louder side than softer. Some recordings can handle being cranked (louder than I would suspect the engineer listened in the control room), yet most sound horrible at higher levels. That is leaving aside things like compression and distortion which become more obvious at higher levels, I'm more talking about the 'rightness' of the sound at a specific level.
I've only just started reading the mouton site, lot's of very good stuff in there. He certainly makes the point about having knowledge of the playback level that should be used as you say, but I'm not sure how realistic that is. He mentions that a good system should have ample dynamic range to handle the the peaks on a symphonic recording or indeed a heavy rock recording. In the case of say heavy metal, the average level would be pretty high, with room for the peaks as mentioned.
Yet he also mentions when mixing that (as they can do it for hours on end) the average level would probably be lower than it would be for lsitening, which tends to overide the notion of 'playing back at the same SPL as the mixing level', yet I feel the earlier thought of it gelling at the right volume definitely applies.
Probably missing something in what he says but hey, that is not unusual!
Re: Re: Re: Why "Flat" is Inaccurate
Do you mean Fletcher and Munson in 1930'th could not have better speakers than they had?
First of all, they had roll off below 100 Hz.
Second, they had roll off above 12 KHz.
Third, they had a resonance around 3 KHz that intuitively was shifted by engineers who designed the speaker because it sounded more natural for male singers than the same resonance on higher or lower frequencies.
So, what F&M measured, they measured how people would correct their sound system so it would sound flatter, i.e. more natural...
...but they were in love with their sound system so decided to blame on experts who performed the blind test... But as soon as they started blaming on well known and well respected experts, they had to generalize and say that all human beings has such natural hearing defect...
CeramicMan said:
Do you mean Fletcher and Munson in 1930'th could not have better speakers than they had?
First of all, they had roll off below 100 Hz.
Second, they had roll off above 12 KHz.
Third, they had a resonance around 3 KHz that intuitively was shifted by engineers who designed the speaker because it sounded more natural for male singers than the same resonance on higher or lower frequencies.
So, what F&M measured, they measured how people would correct their sound system so it would sound flatter, i.e. more natural...
...but they were in love with their sound system so decided to blame on experts who performed the blind test... But as soon as they started blaming on well known and well respected experts, they had to generalize and say that all human beings has such natural hearing defect...
terry j said:
The problem is that every room is different, and speakers do not all have the same off axis response. That makes it hard to give a good rule of thumb.
If you search google for 'room curve' or 'house curve' you'll find something. Here's one graph I had saved.
Rob.
Attachments
Re: Re: Re: Re: Why "Flat" is Inaccurate
Hi,
That wasn't what I was talking about.
IMHO the Fletcher Munson/Equal Loudness Contours are merely a curiosity that don't have much use outside of hearing-aid applications.
If you're at a live concert and you get up and walk further away from the singers, they can't be equalized for your sake, so why should loudspeakers be automatically adjusted for an unknown listening distance?
Cheers,
Wavebourn said:
Hi,
That wasn't what I was talking about.
IMHO the Fletcher Munson/Equal Loudness Contours are merely a curiosity that don't have much use outside of hearing-aid applications.
If you're at a live concert and you get up and walk further away from the singers, they can't be equalized for your sake, so why should loudspeakers be automatically adjusted for an unknown listening distance?
Cheers,
Regarding spatial reproduction, the speakers' in-room response should be flat. Period.
The HRTF should be taken into account on the recording side, which is usually not the case unfortunately.
Citation from the Tube CAD journal:
"It is sad to reflect on early death of Alan Blumlein, the brilliant inventor of stereo. He patented stereo in 1933 and he had an amazingly deep understanding of the principles underlying the actual perception of an acoustic event. For example, he knew that the pan-pot approach used in all modern recording studios is inadequate to the task of moving the perceived position of a sound without smearing the sound, as there must also be a frequency selective compensation. "
Regards
Charles
The HRTF should be taken into account on the recording side, which is usually not the case unfortunately.
Citation from the Tube CAD journal:
"It is sad to reflect on early death of Alan Blumlein, the brilliant inventor of stereo. He patented stereo in 1933 and he had an amazingly deep understanding of the principles underlying the actual perception of an acoustic event. For example, he knew that the pan-pot approach used in all modern recording studios is inadequate to the task of moving the perceived position of a sound without smearing the sound, as there must also be a frequency selective compensation. "
Regards
Charles
phase_accurate said:
thanks Charles
just to make sure we are on the same page, if I measured the speakers response at the listening position, it would be a flat line from twenty to twenty??
I'm not particularly disputing it, but speakers that I've measured flat on axis at say a meter always show a dropping response as we go higher in frequency once the room interaction takes place.
It is hard to find concrete data/opinions on this, as suggested earlier did a google on house curve/room curve, and only a few 'usable' results came up.
One that was quite specific was from Accuphase when discussing the dg-28
"If adjustment is performed so as to achieve fully flat frequency response aT THE RGULAR LISTENING POSITION, THE AUDIBLE RESULT WILL BE FELT (sorry bout caps lock) to be very bright. Normally a gradually declining curve towards the upper frequency region is preferrable.
Therefore the dg-28 incorporates preset roll off characteristics, with 1, 2 and 3 db attenuation per octave, starting at 2 khz"
One would presume that is some sort of auto eq function when taken from the listening position.
I have ordered the Alton everst handbook, should get it in a few days.
thanks for chiming in Charles, any more data on what you said????
Just today it happened I calibrated the control room's monitors of a recording studio. The main overhead, in wall Tannoy coaxes, had to be notched in 3 places and get a mild shelve up, from 7k upwards, in order to overcome some overdone absorption in the control room's design. Dips where left alone, since filling them up sounded phasey. The dips are mainly reflection induced by the large monitor for Nuendo and Pro tools in between the coax active speakers, and from the recording room window glass just underneath. They are absent when measuring the nearfield Dynaudios. The measurement of the final 'correct' sounding balance at the place where the engineer's ears are located, has a progressive droop. He used live vocal takes from a Neuman valve mic and mixes of his own, plus Funk, R&B, and Rock tracks. Dacs where Apogee and Benchmark. Mixing and parametric modules are Nuendo.
Attachments
I am a pro, but that one was for a mate. See his nearfield Dynaudio active monitors corrected at his ear's working location. Here we followed a direct sound dominated approach.
''But if you sit quite close to your speakers in a room that is deadened down satisfactorily
(deader is better!), the closeness and deadness will combine to make the diffuse field a small contributor and the direct sound audibly dominant. At the same time, the direct arrival will dominate the picture to the point that the droop in the top will not be very large when the speaker is set to anechoic flatness.''
Robert E. Green - Audio In Modern Times.
- Our best sounding near field control room calibration ended up agreeing with the above.
''But if you sit quite close to your speakers in a room that is deadened down satisfactorily
(deader is better!), the closeness and deadness will combine to make the diffuse field a small contributor and the direct sound audibly dominant. At the same time, the direct arrival will dominate the picture to the point that the droop in the top will not be very large when the speaker is set to anechoic flatness.''
Robert E. Green - Audio In Modern Times.
- Our best sounding near field control room calibration ended up agreeing with the above.
Attachments
panomaniac said:
I am trying to tell that reference headphones, speakers, and microphones are referenced to each other.
So, is "flat" really flat?
Do we really have any way to measure SLP directly, without converting it to movements of some mechanical surfaces that are converted into AC signals?
At least, we have 2 (two) conversions that are frequency dependent by definition.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.