I set it at U87 a while back, will go back into the setup to confirm but I suspect whenever I remove batteries to change it reverts to usual flat response mic. So the sound clips May after all, be recorded with the default flat mic profile as I have changed batteries many times.
U87 is not just a vocal mic I don't think. They use it for instruments and even orchestras.
That could well be a useful data point. Chris is going to build a set of A7.3 supraBaffles for the FH3 this weekend. Only issue is that i only have A7.3eN and they are already notibly better than a stock set in the same room.
dave
There's no winning with any mic with you is there? Frankly the music clip mics are what they are - when we listen to records they are not recorded with calibrated mics. No one does that.
There is no winning with any mic. It is like going from 8 lanes to 2 (and that is likely giving the benefit of the doubt)
dave
dave
There is no winning with any speakers either, hence why we are trying to go to some effort to determine which ones are generally considered good and which are not. A test which provides a fair comparison between a set of drivers is better than no test at all. With a mediocre mic, an awful speaker will still sound awful compared to a great speaker.
Not everyone is going to buy a set of speakers just because one guy on the internet said they were good. Nor do we all have the budget to buy every possible speaker and evaluate them in person, or access to a place where we can try before we buy.
Last edited:
But it is not truly a test between a set of drivers. It is a test between a bunch of drivers convolved with a whole bunch of other stuff that degrades the performance of them all such that no real conclusion can be made. Only the grossest attributes survive, all of the subtle, very important stuff is wiped out.
dave
I didn't say that the gross details did not count, you put words in my mouth. The subtle capabilities are very important. The difference between good & great is the subtle stuff, this test removes enuff information that you cannot even tell that.
dave
dave
I embrace scientific theory. I am only here pointing out the shortcomings of this test… an important part of any scientific test. Ignoring them is very unscientific.
dave
dave
So what are the shortcomings then, scientifically? Skepticism over plastic capsules isn't scientific proof.
Plastic capsules are only a potential problem. A minor one.
To paraphrase from post #2 correcting for new information:
When listening directly to a speaker under test the listener is listening to his system, in his room with the DUT speaker.
In this test the person listening is using his same system and room + his speaker to listen to "the driver being "tested", high passed (removing any influence from the LF -- not good if you want to use the FR as a FR, and adding something from the woofer & the XO topology chosen) convolved at a minimum with XRK's playback system, room, a miniDSP, amp, source, etc, an inexpensive mic/recorder with unknown frequency response & dynamic capability, etc and the losses inherent in the act of recording it, converted to a lossy format".
We know little of XRK's room, and from what i understand, XRK's system is composed of bits with modest capabilities.
The DUT is no longer the speaker of interest.
dave
To paraphrase from post #2 correcting for new information:
When listening directly to a speaker under test the listener is listening to his system, in his room with the DUT speaker.
In this test the person listening is using his same system and room + his speaker to listen to "the driver being "tested", high passed (removing any influence from the LF -- not good if you want to use the FR as a FR, and adding something from the woofer & the XO topology chosen) convolved at a minimum with XRK's playback system, room, a miniDSP, amp, source, etc, an inexpensive mic/recorder with unknown frequency response & dynamic capability, etc and the losses inherent in the act of recording it, converted to a lossy format".
We know little of XRK's room, and from what i understand, XRK's system is composed of bits with modest capabilities.
The DUT is no longer the speaker of interest.
dave
I think you are jumping too quickly to the conclusion that distortion in the measurement is simply just 'throwing information away'. It's not like that. The only information that is truely 'thrown away' is any information which is below the noise floor of the measurement system. Linear and non-linear distortion will add noise to the measurement but not take anything away.
Obviously the person conducting the test must take adequate measures to reduce the noise and distortion floor of the system as much as possible. This includes noise+distortion floor of the DAC, speaker amplifier, testing environment, mic, mic preamp/phantom power/ADC.
I have already pointed out that my low budget setup produces a 90-100dB noise floor at 96dB SPL (at or below the threshold of human audibility) and only a 2nd order harmonic at -50dB. A fair compromise to most.
The non-linear distortion of the mic is a pretty common issue. Behringer ECM8000 does exactly the same thing. It is my understanding that modified Panasonic WM-61A capsules avoid the 2nd order distortion and I would assume many high end mics do as well.
Linear distortion is typically going to only be limited to +/-1dB, maybe 2dB if your mic cal is exceptionally inaccurate, or your mic cal was done at 25*C and your ran the test in a room at 40*C. Lets not forget that the speaker drivers themselves are also subject to linear distortion due to temperature, humidity, inappropriate storage methods and so on. If I buy an "EnABLed" speaker from yourself, who's to say it will arrive 'in calibration' after being shipped halfway around the world?
Even if the test setup introduces an audible amount of linear distortion, at least that is the one thing you can change in a speaker system with almost no compromise. If I listen to a speaker in one of these 'tests' and decided to buy it only to find it is a bit hot in the 5-8KHz range it is easy enough to shape the response back down with EQ or passive filtering. On the other hand if you buy a driver and it has horrible non-linear distortion, you are pretty much stuck with it. The noise and distortion floor of my test setup is appropriate that a driver with horrible non-linear distortion is plainly audible.
Lastly, we consider the ability of the end users to reproduce what was recorded faithfully. Their own speaker/headphone systems will add noise and distortion. Thankfully, most end users will have mentally adapted to their own system establishing a reasonable baseline of what is 'neutral'.
Obviously the person conducting the test must take adequate measures to reduce the noise and distortion floor of the system as much as possible. This includes noise+distortion floor of the DAC, speaker amplifier, testing environment, mic, mic preamp/phantom power/ADC.
I have already pointed out that my low budget setup produces a 90-100dB noise floor at 96dB SPL (at or below the threshold of human audibility) and only a 2nd order harmonic at -50dB. A fair compromise to most.
The non-linear distortion of the mic is a pretty common issue. Behringer ECM8000 does exactly the same thing. It is my understanding that modified Panasonic WM-61A capsules avoid the 2nd order distortion and I would assume many high end mics do as well.
Linear distortion is typically going to only be limited to +/-1dB, maybe 2dB if your mic cal is exceptionally inaccurate, or your mic cal was done at 25*C and your ran the test in a room at 40*C. Lets not forget that the speaker drivers themselves are also subject to linear distortion due to temperature, humidity, inappropriate storage methods and so on. If I buy an "EnABLed" speaker from yourself, who's to say it will arrive 'in calibration' after being shipped halfway around the world?
Even if the test setup introduces an audible amount of linear distortion, at least that is the one thing you can change in a speaker system with almost no compromise. If I listen to a speaker in one of these 'tests' and decided to buy it only to find it is a bit hot in the 5-8KHz range it is easy enough to shape the response back down with EQ or passive filtering. On the other hand if you buy a driver and it has horrible non-linear distortion, you are pretty much stuck with it. The noise and distortion floor of my test setup is appropriate that a driver with horrible non-linear distortion is plainly audible.
Lastly, we consider the ability of the end users to reproduce what was recorded faithfully. Their own speaker/headphone systems will add noise and distortion. Thankfully, most end users will have mentally adapted to their own system establishing a reasonable baseline of what is 'neutral'.
Last edited:
Huh? All devices have information loss which measured distortion does not give us any clue about. And that we are listening to X's room adds and hides and the act of recording creates large amounts of info loss.
All important, but irrelevant to my points.
Next to meaningless information. It is possible to have many systems that can produce those same numbers and sound very different.
You talk of distortion. We have not yet figured out how to measure distortion (in the global sense). We do have tests, but as used the data is meaningless, and the data that isn't (maybe) has yet to be scientifically correlated with what the ear/brain perceives (there are but a few tidbits*). http://www.diyaudio.com/forums/ever...d-meaningless-harmonic-data-what-matters.html
You seem to have a very simplistic (and not very good) grasp of what a hifi does.
dave
*(one being Geddes' very scientific test which points to the distortion measures we make on speakers being meaningless)
What you are describing is not explained by any scientific theory to the best of my knowledge.
As far as i'm aware, the recording system will be characterised by a transfer function which introduces linear and non-linear distortion in addition to random noise with some frequency distribution. The only information that is truely 'lost' and unrecoverable is the information which is smaller in amplitude than the random noise. If the noise floor is below human audibility, who cares that something that was not audible becomes buried in the noise floor of the measurement?
90-100dB Noise floor isn't audible. -50dB second order harmonic is something we'd like to avoid, but subjectively isn't very intrusive. Any other system which produces the same numbers of noise and non-linear distortion vs signal amplitude will sound pretty much identical. No one has scientifically proven that phase distortion (and therefore the phase of the harmonic) is audible within a band of more than 1/3rd of an octave, and a 2nd harmonic places it at 1 octave to the fundamental.
For comparison, a speaker driver e.g. Fostex FF85WK introduces a 2nd order harmonic at -45dB, 3rd order at -50dB and 5th order at -70dB. This is much more subjectively intrusive or imparts 'loss' on the recording as you put it.
You just answered your own question. The thread you linked to points out that a generalised THD figure does not describe a system, but individual quantities of each harmonic do. A "5% THD" system may sound very different to another "5% THD" system as one system may be 5% 2nd harmonic and the other may be 5% 10th harmonic.
A system that is 2% 2nd harmonic, 1% 3rd harmonic and 0.1% 5th harmonic should sound impart the same audible harmonic distortion as any other system that is 2% 2nd harmonic, 1% 3rd harmonic and 0.1% 5th harmonic. Obviously there is a relationship between signal amplitude and harmonic content, but the transfer functions of speakers and microphones especially are very comparible because they all tend to suffer from the exact same mechanisms of non-linear distortion. Therefore they all tend to follow a similar distortion vs signal amplitude curve. I.e. if you test two speakers at 90dB SPL and one is exceedingly better, it will probably also be exceedingly better at 80dB and 100dB SPL.
Last edited:
non-linear distortion = additional noise, additional noise = worse signal to noise ratio = bad. There is more than enough scientific evidence out there which relates harmonic distortion to a degradation of sound. What has not been scientifically tested in depth to the best of my knowledge is the audibility of the order of harmonic distortion, but it is well accepted that the higher the order, the more audible or unpleasant.
This is in comparision to your theory of "loss" which is based on no measurable parameter or scientific theory, merely your own gut feeling.
Ain't that the truth.
Last edited:
- Status
- Not open for further replies.