[vergesslich2]

Hi : )

I wondered if anyone could quantify (long term) influences of sound systems like speakers and headphones on productions and thus on other headphones? When I try to answer it myself I notice that I have not enough knowledge.

E.g. for a while Yamaha NS-10 were popular, they had a precise impulse response and could sound somewhat bright (correct?), so the producers might have (did they?) created productions with less treble. Then other manufacturers might have found a reason in this to make their speakers like NS-10 (or otherwise bright) instead of doing this for the scientific reason that a good impulse response should give a more exact sound reproduction. How would I know which reason is true?

Have there been enough things to experiment with so that the sound reproductions follows the scientific (versus economic) reasoning? E.g. reproducing the human voice and listening to it and trying to blind-test if there is a person in front of you or not, or something like that maybe?

I thought, well, the whole thing would regulate itself either way. If there were influences of systems on productions, and then adaption by other manufacturers, it wouldn't really matter, the progress in this field cannot diverge from a psycho acoustic reasoning much, because otherwise you would hear it. But is that true? I have no idea, what the economic reasoning is, what the scientific reasoning is, and what actually neutral (frequency) response would be, or any orientation at all.

Another question:
Can there be accidental analog dithering? Can a noise floor in gear make impulses or highs make sound smoother? In discrete / digital systems with bits and bytes, one can use dithering by adding specially crafted "noise", and in a DAW I found this to make a channel sound somewhat better or smoother. Does this mean, that in the analog world, a noise floor could somehow achieve the same? I guess no, because it doesn't fit my understanding of dithering (to improve the sound of low bit-depth). But if someone could explain why not, this would be insightful.

Kind regards

 

[bigshot]

Production studios have custom designed rooms and calibrated speaker systems. They don't depend on the way the speaker itself sounds. They do room treatment and equalization to optimize the playback to a target curve. That way, if you start a mix in New York and finish it in Los Angeles, it will sound the same in both studios. The particular brands and models of speakers favored by recording studios are preferred because they are flexible enough to be calibrated from 20 to 20, not because they sound good out of the box.

 

[vergesslich2]

Thanks : )

The thing I tried to understand is how the studio got a perfect microphone for the calibration in the first place. Are there microphone brands that try to reject any influence from unscientific (e.g. random music) productions? I guessed so. And I guessed they do so with a speaker that is made under the same premise. But how do they get that speaker? So at least someone had to approximate and improve on and on their products, using a human to do a listening test with something a human can recognize easily, in order to tell if the sound reproduction was successful. So it's a psycho acoustic way to get to anywhere at all, is it? Or is there a purely physical way, not involving listening tests?

I got interested in it because I read, that the Harman Curves are based on comparisons with speakers. I assume it's done like this because flat speakers in treated rooms exist, and that's the starting point. But that would mean, flat headphones are based on flat speakers instead of the above approximation procedure with headphones and microphones (or the purely physical way). Are there Harman Curves that did not involve speakers?

Sorry for my preconceptions in my question. It's the way I try to get rid of the non-facts in my head. I have no big picture how people got to their measurement equipment in the first place. I was bad in physics.

 

[bigshot]

They calibrate using test tones. The frequency tone is played back in the room and measured with a very accurate calibration microphone. Then they balance the EQ bit by bit throughout the entire audible range. Once the room is designed and treated, they can get very accurate results. It’s a matter of measuring the amplitude of each frequency band. It doesn’t involve listening. It’s more precise than that.

The Harman curve is not based on speakers. It’s based on polling people for their preference in headphone sound. Speakers are a whole different animal. You can’t make headphones sound like speakers without doing sophisticated headphones tracking, calibration to each individual listener, and complex signal processing. Response is only a small part of it.

 

[vergesslich2]

The Harman curve is not based on speakers. It’s based on polling people for their preference in headphone sound. Speakers are a whole different animal. You can’t make headphones sound like speakers without doing sophisticated headphones tracking, calibration to each individual listener, and complex signal processing. Response is only a small part of it.

Oh, true. That ("based on speakers") was a thing I just googled before. I 'knew' it was about polling, googled, suddenly I 'knew' it would be about speakers, because some text mentioned a speaker. Didn't really read it.

https://www.headphonesty.com/2020/04/harman-target-curves-part-3/

This text mentions that there are some derivations in Harmans whole work coming from speakers, but nothing about it being based on speakers. I focused on my premise "what could influence it just the slightest" too much.

Is headphones tracking tracking the movement of peoples head? I wondered about that too. I was always a bit suspicious about binaural recordings of noises in rooms (not of headphones, because they move along with your head whereas noise sources in a room relatively stay) or something like that. I'm quite unsettled with my head. Kind of why I thought I don't buy a Nintendo 3DS that wants to track my eyes.

Off-topic (if not already):
Glad I didn't try to prove a point but rather told you I know almost nothing about the matter. ; ) I'm actually looking for a headphone and tried to understand why some Focal headphone was (again judging from a dummy head recording -> to my speakers) a bit better than the hd 650. Although, don't know, maybe it has just the ability to sound like some multi-band-compression tool and I was just amazed what details the recording brought to my speakers.

I think I might understand the whole thing at some point.
Kind regards and thanks

PS: Just one little thing. Normally I'm a bit ignorant towards physics, but once I watched a video about the Cosmic distance ladder (called more like "reference candle" in German) and it stuck in my head how challenging it was (for me to read) how physicians would be able to agree on measurements. The perfect microphone, the reference microphone, must have some cool physics behind it. Because, I'm struggling to understand how something can emit a signal, another thing receives it, and in the end you're able to judge whether the emitter was more faulty than the receiver, or the way round. I think that's the reason for my posting.

 

[castleofargh]

http://seanolive.blogspot.com/2009/10/audios-circle-of-confusion.html
The book Sound reproduction from Floyd Tool is still a reference IMO, because while it's focused on speakers and rooms instead of headphones, it brings together all steps of production and playback to see the problem as a whole. Then he discusses them one by one with the good old(I freestyle paraphrase, sorry) ”listen to the science, not just to popular beliefs”.


Audio is hard, while astronomy is super easy. You just need to master chemistry, spectrometry, nuclear physics, a ”bit” about optic, crazy high level of math, and fully understand space-time to account for gravitational lensing, red shifting and all that. Not a big deal.

The English name is ”standard candle”(same thing in French)

 

[bigshot]

This text mentions that there are some derivations in Harmans whole work coming from speakers, but nothing about it being based on speakers.

Sidney Harman and Bernard Kardon were speaker designers. John Lansing was the speaker designer behind JBL. JBL is now a division of Harman-Kardon. H-K is involved in all kinds of home entertainment products. The Harman Curve was established by the research & development department of H-K, which also researches speaker acoustics and design. So there is a link to speakers, but the Harman Curve really doesn't have anything to do with speakers. It's a headphone curve.

Is headphones tracking tracking the movement of peoples head?

Yes. Currently the Apple AirPods Max and the Smyth Realiser are capable of head tracking. We will probably see more products hit the market in the not to distant future.

I'm actually looking for a headphone and tried to understand why some Focal headphone was (again judging from a dummy head recording -> to my speakers) a bit better than the hd 650.

A dummy head recording being played back on speakers isn't going to give you anything coherent. Speakers use the room to create soundstage, which is the opposite of how headphones work. It's kind of like using a screwdriver to hammer a nail.

I think I might understand the whole thing at some point.
Kind regards and thanks

I watched a video about the Cosmic distance ladder (called more like "reference candle" in German) and it stuck in my head how challenging it was (for me to read) how physicians would be able to agree on measurements. The perfect microphone, the reference microphone, must have some cool physics behind it. Because, I'm struggling to understand how something can emit a signal, another thing receives it, and in the end you're able to judge whether the emitter was more faulty than the receiver, or the way round. I think that's the reason for my posting.

You're asking the wrong person about "cosmic distance ladders"! But as for your last sentence... If you have a target curve, meaning the coloration you want the sound to have, then you can compare your measurement of the output to the target curve and see if the output needs a little more bass or a little less treble. You go through frequency band by frequency band measuring the output from the speakers and use an equalizer to conform the output to the target curve. Does that make sense?

 

[sander99]

The thing I tried to understand is how the studio got a perfect microphone for the calibration in the first place.

They calibrate using test tones. The frequency tone is played back in the room and measured with a very accurate calibration microphone.

I have a feeling @vergesslich2 is hinting at a more fundamental problem that must have been solved at some point in the past, one that I also don't have the answer to:
To do a very precise calibration you would either need a microphone with a very flat frequency response or a microphone of which the frequency response is very precisely known. How did the first of such microphones come into existance? How do you determine very precisely the frequency response of a microphone, or determine that a microphone has a very flat frequency response if you don't have another one such microphone (flat or with a known frequency response) already to compare against, or a speaker with a very flat or precisely known frequency response? And if there was already such a speaker how would you know without already having such a microphone?
Chicken or egg...

 

[bigshot]

There have been calibration mikes for many decades. And there have been meters to measure amplitude for just as long. If you want to know how the "flat" target curve was determined, you would have to go back to the studies at Bell Labs in the 1920s. That's where the aspects of audio fidelity were defined. I'm guessing they measured real sound and compared it to recorded sound and compared that to the balances of human hearing to determine their target curve.

 

[ADUHF]

I have a feeling @vergesslich2 is hinting at a more fundamental problem that must have been solved at some point in the past, one that I also don't have the answer to:
To do a very precise calibration you would either need a microphone with a very flat frequency response or a microphone of which the frequency response is very precisely known. How did the first of such microphones come into existance? How do you determine very precisely the frequency response of a microphone, or determine that a microphone has a very flat frequency response if you don't have another one such microphone (flat or with a known frequency response) already to compare against, or a speaker with a very flat or precisely known frequency response? And if there was already such a speaker how would you know without already having such a microphone?
Chicken or egg...

This is not a subject I'm very familiar with. But I believe there are fairly rigorous international standards for the adjustment and calibration of microphones used for precise scientific and acoustic measurements which are performed by national measurement institutes, such as NIST and The National Physical Laboraty in the UK.

It looks like the favored method used by these organizations involves something called "reciprocity calibration", which is explained in some detail here...

https://en.wikipedia.org/wiki/Measurement_microphone_calibration

"The technique provides a measurement of the sensitivity of a microphone without the need for comparison with another previously calibrated microphone, and is instead traceable to reference electrical quantities such as volts and ohms, as well as length, mass and time."

The above article references:
IEC 61094-2, edition 2. (February 20, 2009) "Measurement Microphones, part 2". IEC Standard for Pressure Reciprocity Calibration of Measurement Microphones
IEC 61094-5, edition 1. (October 16, 2001) "Measurement Microphones, part 5". IEC Standard for Comparison Calibration of Measurement Microphones

 

[ADUHF]

This looks a bit dated (the most recent citation is from 1992), but here is a NIST white paper on the above...

https://www.nist.gov/system/files/documents/calibrations/aip-ch8.pdf

A reciprocity calibration system from the NIST website...

 

[vergesslich2]

http://seanolive.blogspot.com/2009/10/audios-circle-of-confusion.html
The book Sound reproduction from Floyd Tool is still a reference IMO, [...]

Thanks for hinting me to this. I'm just reading it, got to page 15, but as the foreword said, reading it linearly might be not exactly right, and I'm a especially slow reader, re-reading parts... anyway, the book is *so interesting. The audio's circle of confusion is exactly what made me so confused. ; )

That the first or one of the first listening tests took place in 1918 at the Edison studios is what I somewhat subconsciously anticipated, but there's not much popular documentations about such things, so... I would never have thought that Edison might have cheated a little bit by making Frieda Hempel sing like the loudspeaker sounds.

 

[vergesslich2]

You go through frequency band by frequency band measuring the output from the speakers and use an equalizer to conform the output to the target curve. Does that make sense?

Yes. I have more of a computer background and can understand the divide and conquer approach.

I'm reading Sound Reproduction by Floyd Toole, so I might in the end even know some numbers and be enlightened a little bit, in order to eventually make a good headphones buying decision, and for musical purposes in general.

 

[bigshot]

I would never have thought that Edison might have cheated a little bit by making Frieda Hempel sing like the loudspeaker sounds.

Actually, they selected singers whose range sat comfortably within the sweet spots of the horn. The first recording star was Billy Murray and his voice exactly nailed the frequency response and wolf tones of the Victor horn. (They didn't have loudspeakers back then. Recording was acoustic, no electricity involved.)

 

[vergesslich2]

This is not a subject I'm very familiar with. [...]

https://en.wikipedia.org/wiki/Measurement_microphone_calibration

Unfortunately I'm not familiar with physics at all. But it's fun reading about it, to be honest; and it's kind of fun to wonder whether it could have something in common with Kalman filters... well, if not, I should accept that one would have to understand each thing separately... in any case, thanks. : )