It was incorporated into Praxis, but required Praxis' hardware "Audpod" adaptor/key (off the market for a long time now). It's difficult to keep a product viable when your main competition is well-executed freeware!
Member
Joined 2003
Main benefit is being able to provide some real analysis using a test signal that approximates real world audio, or simply is real world audio. No longer can people criticize that the performance measurement does not represent real world audio, since sine wave is only 3dB crest factor, where real world audio is 12dB+. But as I've seen in this thread, the people have found all sorts of new ways to criticize, so that will never end.
Potential value add of real world crest factor stimulus is not needing some high SPL sine wave stress test just to see what a speaker may do for the occasional peak in output, for example a common HD sine sweep test is at 86dB and 96dB/1m. There's also the common misconception that the HD result of 86dB SPL is representative of performance at 86dB RMS audio level, which is isn't, simply due to the difference in crest factor.
Another value add for REW, forgetting about any distortion analysis, is the ability to generate an accurate impulse response / frequency response from any audio. You don't have to listen to that annoying sweep anymore just for a frequency response.
The ability to separate the signal from distortion and listen for yourself is an added bonus, as Bill Waslo mentioned, not truly a new feature but now is available free to all DIYer in modern, actively supported and developed software.
Last edited:
We know all these methods are OK for response provided THD isn't too 'large' ... including using music as Bill Waslo and I have done in da previous Millenium
Can we please call 'ESS' its proper name; Angelo Farina's method. It's actually a logarithmic (not exponential) sweep as I've argued with Prof Farina on many an occasion.
It's the sweep you get from a B&K 2010 driven by a 2307 chart recorder; response as God intended 😊 I no longer have these as it's impossible to get the steam to run them. I do have B&K toilet roll stolen from Naerum.
It's Angelo's simple (??) but powerful processing that makes this the theoretically fastest way to a given accuracy in the presence of noise.
I'm certainly not criticizing you or sm52. Just pointing out that NO ONE has managed to get a FSAF THD measurement that sorta aligns with THD done with Angelo's method. Remember, FSAF is OK for response.
There are several methods which allow the 'modelled' response to be compared the real thing ie what FSAF claims to do. Bill has done so commercially and I've done so experimentally in da previous Millenium as computing power got rapidly cheaper.
But IMHO, these illustrate the non-Time Invariant (compression) aspects, rather than the Linear (distortion) aspects of our assumption that speakers are LTI. So far, the examples I've listened to bear this out ... but as they aren't the result of FSAF Total Distortion measurements that even slightly resemble more traditional methods, I'm not sure they are representative of what is actually happening.
https://www.roomeqwizard.com/betahelp/help_en-GB/html/fsafmeasurement.html
USB microphones which rely on the USB clock are not suitable for FSAF measurement as the USB clock is not stable. That means, for example, UMIK-1, UMM-6 and Omnimic are unsuitable for FSAF measurements. It is possible to use a USB mic which has an internal clock source, such as UMIK-2.
I have only UMIK-1 mk2
USB microphones which rely on the USB clock are not suitable for FSAF measurement as the USB clock is not stable. That means, for example, UMIK-1, UMM-6 and Omnimic are unsuitable for FSAF measurements. It is possible to use a USB mic which has an internal clock source, such as UMIK-2.
I have only UMIK-1 mk2
I thought I'd answered this recently but it must have been in another thread.
All these methods are for weakly non-linear systems. ie they assume THD is 'small'. If this is so, they all give 'similar' response with different accuracies and efficiencies.
When THD is high, eg when the device / signal clips, this breaks down and you get very different results. Angelo's method is probably the best at (not) reacting to this and MLS the worst. MLS can't measure distortion at all and is seriously affected by it. Dunno about FSAF and it's one of the things I want to find out
DcibeL, can you confirm that the test signal for both Angelo's method and FSAF in #58 for the two speakers are exactly the same? ie The same 'exponential' sweep, the same start & end, same length & level.
This is what hifijim suggested 😊
If this is the case, we can get some important results and conclusions.
This is what hifijim suggested 😊
If this is the case, we can get some important results and conclusions.
Missed this. Could you repeat the FSAF test on that speaker using the same 'exponential' sweep. This would give us the closest conditions for the two methods.
The Dynamic Duo, Lipshitz & Vanderkooy who helped develop MLS in Jurassic times, would think us very silly to try to measure HD with MLS ... but I'm sure they would be delighted to be proven wrong 😊
Last edited:
But MLS is better for noisy environment. You also need to keep in mind, measurement results obtained using the sweep sine technique can vary dependent on the sweep rate, which is known problem in vibroacoustic measurements. For instance, observant experimenters have probably noticed that when measuring impedance with a sweep tone in REW, the shape of the impedance curve depends on the sweep tone rate.
In the past measurements were developed to prove a theory. Today measurements are quoted to substantiate righteousness. I know several professionals (non-engineers) who own substantial hi-fi systems purchased by listening only without having a clue what is discussed here.
Digital show and tell from Monte Montgomery, and the geniuses who gave us FLAC
https://xiph.org/video/vid2.shtml
Also:
Measurements are hard
Measurements are not hard, but your measurement technique and interpretations with a knowledge of error posibility to obtain accurate measurements are. I think few people know what the mirror is for on an analogue meter scale.
Omnimic has an internal crystal clock source, change the capacitors across the crystal, the rate changes slightly.
I wonder if the statement refers to inability of single input channel sources to sample synchronize with output DAC clocks, which prevents some measurement methods.
Sam Berkow's SMAART software was measuring frequency and impulse response using music or noise in 1996. It works well given time, though dynamic range with such signals isn't optimum and the program material needs to have sufficient energy across the measured spectrum Measuring distortion with it would seem to be FSAF"s unique ability though.
Member
Joined 2003
Sure, give me some time, maybe tonight or tomorrow.
Is anyone in this thread besides myself and tktran going to open REW and push the measure button?
Stepped sine wave is best for a noisy environment. Great out of band suppression of noise. Very annoying, not fast. Tough on the device under test. But very robust when you have the time and the ability to wear hearing protection.
Looking into it yes. Perhaps this week. I have a lot of Reading the manual to do.
Member
Joined 2003
I think you mean worse. MLS or noise signal is exciting a wide spectrum for the entire duration of the measurement, which allows for greater potential for contamination of background noise. Sine Sweep is only observed a single frequency for a given point in time, so there's less opportunity for background noise to contaminate the measurement.
In your own citation provided above, it indicates higher SNR of a sine sweep vs MLS/noise signal..
Anyway, let's try to steer back on topic and not get too detailed by semantics.