That explain why several time, listening to a "golden star" enclosure, i was so deeply disappointed at the first second ?
About ears/measurement, i believe it would be impossible for me to build my enclosures without a Neutrik, but i had spend months to tune the level of the horn at 1/4dB, no differences on the curves, big difference with ears, as it is half the energy/bandwidth..
It is a common practice, during mixing session, to add some 40KHz peak bump to a female voice, to give some air, and everybody can feel the difference, even few dBs. And i'm not able to ear 20Khz pure sinus since a long time.
Last edited:
I see only an issue when calibrating speakers only with measurements - the ear is not a microphone 
Harman is one of the few if not only manufacturer that takes listening tests very seriously, and uses a panel of trained listeners, blind tests and has developed some good software to teach how to listen.
Link to a public beta is on Sam Olive blog (I'm on phone and don't have link at hand but google will do)
Harman is one of the few if not only manufacturer that takes listening tests very seriously, and uses a panel of trained listeners, blind tests and has developed some good software to teach how to listen.
Link to a public beta is on Sam Olive blog (I'm on phone and don't have link at hand but google will do)
Last edited:
Getting the leading edge of the transients right is essential, but most systems distort too much on playback to sound "correct". Nothing wrong with the recordings, the failure is on the end consumer's side - in the effort to not have the reproduction too abrasive for a high percentage of recordings the sound is so dulled or "politisized" that very little playback ever hits the mark ...
BTW, Merry Christmas to all! ...
Frank
I've done some of the brickwall filtering tests myself, I personally can't hear a 17kHz brickwall applied to music (IIRC a pretty nice set of Prokofiev string concertos) so I don't bother being a subject in these tests. But I thought we started talking about folks who could not hear anything but their brains were working on it.
I've spent years designing experiments for problematic circuits and customer applications and acoustics is very problematic, the quality of instrumentation for sound pressure is so far behind that for voltage and current (for instance) that the room for error is large. LP's, speakers, and microphones are orders of magnitude worse for IM and FM than precision electronic instrumentation.
Happy holidays to all, I manged to get all but one kid and my buddy Otis together here in SF.
Last edited:
I can second that. My little contribution to BlindLeaks is that the top speaker designers in the known universe all have some hearing affliction or other.
But they can all lip-read speakers behind a black curtain.
I'm not sure the usual hearing loss with age precludes a subject from being a true golden pinnae except for severe cases and for special tests like audibility of supersonics.
"Professional audio reviewers" with one or two notable exceptions perform badly compared to the general public. Jane Public is significantly better than Joe. But the consistency of responses from the public is remarkable. The key is removing stress.
We never did hearing tests on subjects BEFORE a listening test but would test those who did well in the Blind Listening Tests. Our 'formal' panel was tested at least once a year.
Wharfedale did this from the late 70's. The Pseudo Prophets Floyd & Olive quote Peter Fryer in their pontifications. I simplified, refined and improved these tests over nearly 2 decades and in the process was privileged to work with some of the best ears in the business.
Excuse my obvious NIH syndrome when discussing Floyd/Olive. Since Wharfedale became a box stuffer, Harman are probably the only maker attempting to use Blind Listening Tests as a design tool today so deserve recognition. Most of their stuff is good though I can laugh at some of their methods.
Standing back as far as I can, my main criticisms are
- not paying enough attention to reducing stress. I only know what they have published so they may have addressed this
- their tests are obviously not designed by speaker designers
This was done circa 1980 with vinyl & mastertapes we cadged off the record companies.
One of the recordings was the famous early Blumlein stereo Beecham Scheherazade from EMI. I was always unimpressed by the vinyl but the mastertape was a revelation.
Bandlimiting was by brickwall passive analogue filters at 20kHz; essentially souped up multiplex filters. In fact the first trials were with multiplex filters screwed up to their limit. Peter Wall could do supa dupa passive filters the old fashioned way. I grovel at this feet.
As a speaker & microphone man, I can pontificate at length on the reasons for the preference for band limiting but the intermod that Scott & others have pointed out is probably the main cause.
The reason for these tests was prompted by some inconsistency in doing other audibility tests.
In the 90's, we considered repeating the tests with better source material but by then, all high quality sources were evil digital and already bandlimited. This makes life a lot easier for playback electronics.
The results are exactly applicable to Golden Pinnae today with their MC cartridges & vinyl ... but I'll stick my neck out and suggest that if these tests were repeated on true SACD (no evil PCM masters issued on SACD), the preference will still be for Bandlimiting.
The issue isn't whether certain people can hear supersonics. It's easy (??) to devise test signals that allow certain individuals to detect supersonic signals ... and leave them with a headache for the next 24hrs.
The issue is whether they can be heard on music. And if the answer is yes, do these listeners prefer their music with supersonics?
To answer these two questions, the tests MUST be Double Blind as the expectation bias in both camps is extremely high.
But the definitive lesson on how a Bandwidth Limitation Test should be conducted is from Harry O, The experiment that saved high fidelity
In the forseeable future, it's likely Bandwidth Limitation will be preferred by those who can tell the difference ... until us evil speaker & microphone designers get our act together
Last edited:
As documented somewhere in the NL AES section at LinkedIn we did some controlled listening tests with the 15kHz brick wall filters used in FM stereo multiplex encoders.
Out of 4 listeners, 3 could hear differences at 100% reliability with/without the filtering (the fourth person stated he could not hear it and did not participate further).
All three we preferred the music with the filter...
Merry Christmas, Happy 2013 to all, and here's some live music for you.
jan
I don't necessarily agree with bandwidth limitation as being a good thing.
I do agree that with MARGINAL digital reproduction, bandwidth limitation is probably a GOOD THING, something that Dr. Diamond found to be true about 30 years ago, with Sony digital playback.
The extension of audio bandwidth is NOT that great a problem, in practicality, because most USEFUL and popular microphones bandwidth limit below 40KHz for sure, and usually 20KHz or so, generally. It is the same with phono playback, analog tape playback, and digital playback as well.
I found, decades ago that the TIM 30 signal is about the highest reasonable worst case risetime (risetime being generally related to bandwidth) of 10uS, is about ALL anybody can do, realistically, and this only with certain musical instruments. IT JUST ISN'T THERE, FOLKS! Now, what about DISTORTION? YES, we can have all kinds of distortion and digital artifacts in the ultrasonic region. That is what is probably being detected, and it SHOULD be removed if possible.
For Me to put a severe low pass filter into my amps or preamps is laughable. I just have to make them fast enough to pass and not distort anything coming through them, and that is what I do.
I do agree that with MARGINAL digital reproduction, bandwidth limitation is probably a GOOD THING, something that Dr. Diamond found to be true about 30 years ago, with Sony digital playback.
The extension of audio bandwidth is NOT that great a problem, in practicality, because most USEFUL and popular microphones bandwidth limit below 40KHz for sure, and usually 20KHz or so, generally. It is the same with phono playback, analog tape playback, and digital playback as well.
I found, decades ago that the TIM 30 signal is about the highest reasonable worst case risetime (risetime being generally related to bandwidth) of 10uS, is about ALL anybody can do, realistically, and this only with certain musical instruments. IT JUST ISN'T THERE, FOLKS! Now, what about DISTORTION? YES, we can have all kinds of distortion and digital artifacts in the ultrasonic region. That is what is probably being detected, and it SHOULD be removed if possible.
For Me to put a severe low pass filter into my amps or preamps is laughable. I just have to make them fast enough to pass and not distort anything coming through them, and that is what I do.
Last edited:
I am of the opinion that the stress that is directed at all this high frequency content is due to high frequency driver distortion and not to a simple phenomena of human hearing and a predilection to dislike of ultra sonic sounds. I have seen very few devices that would not have breakup modes in the upper frequencies and most will have resonances below 20Khz. This is what I think is being detected by the ear. Any frequencies produced above the resonant frequency will excite the device and produce that distortion that you are reacting to. It does not relate to the musical content and this is the distress as far as I can understand it. I have listened to compression drivers and dome tweeters and most will have this affect. The brick walled filters are going to limit this problem but any device that has that type of upper frequency problem can be excited by other frequencies above or below resonance. My Gold Standard at this time is the TAD ET-703 compression driver. This is the standard that I am using for the development of a new dome tweeter. If I can equal this performance in a dome tweeter I will be more than satisfied. And no I do not expect the efficiency of a horn loaded device. I have heard no other device with this resolution and lack of distressful upper frequency response.
First of all, Merry Christmas John!
It was in the original music*. Most is lost with the microphones, as you rightfully said. I know of only a few Bruel & Kajer mics capable of 100khz. I wonder what they use to make DXD masters. The ones from 2L are AWESOME. If you hear them not resampled nor filtered, everything else will sound like reproduced sound.
(* There's life above 20 kilohertz! A survey of musical instrument spectra to 102.4 kHz)
Or pushed much higher in the ultrasonics, where it REALLY can't be "heard".
The approach i'm following is (always integer) upsampling to DXD or higher rate and no filtering.
I'm not sure how capable should be the tweeters, but I wont settle for less than 50khz.
It was in the original music*. Most is lost with the microphones, as you rightfully said. I know of only a few Bruel & Kajer mics capable of 100khz. I wonder what they use to make DXD masters. The ones from 2L are AWESOME. If you hear them not resampled nor filtered, everything else will sound like reproduced sound.
(* There's life above 20 kilohertz! A survey of musical instrument spectra to 102.4 kHz)
Or pushed much higher in the ultrasonics, where it REALLY can't be "heard".
The approach i'm following is (always integer) upsampling to DXD or higher rate and no filtering.
I'm not sure how capable should be the tweeters, but I wont settle for less than 50khz.
Telstar,
Couldn't agree with you more that not much can compare to a B&K mic for high frequency response. The only thing to use for loudspeaker testing above 20Khz with any real usable data. Not sure that there are really many devices that can reproduce above 20Khz and still be listenable though. Just a few that I would even consider to try and do that.
Merry Christmas to all of you in Europe.
Couldn't agree with you more that not much can compare to a B&K mic for high frequency response. The only thing to use for loudspeaker testing above 20Khz with any real usable data. Not sure that there are really many devices that can reproduce above 20Khz and still be listenable though. Just a few that I would even consider to try and do that.
Merry Christmas to all of you in Europe.
- Status
- Not open for further replies.