Q7: Yeah, but when we use a higher bandwidth measurement like 45 kHz or 90 kHz we can see the THD of high frequencies "ballooning" with high distortion. These manufacturers are lying by using using lower bandwidth 20 kHz to hide them! What about the measurements standards from last century?
A7: Remember that this is THD, meaning the harmonics of 10 kHz are 20 kHz, 30 kHz, etc, and the harmonics of 15 kHz are 30 kHz, 45 kHz, etc. They are all inaudible. This is intentional by design: maximize feedback loop to reduce distortions where it is audible below 20 kHz, and let it go where it is inaudible. See
here. The 20 kHz bandwidth measurements just represent better what we actually hear.
Thank you for this update.
Here's the thing. We have a couple posters on ASR hounding us relentlessly with the supposedly critical importance of measuring SINAD at bandwidths well above 20 kHz, even up to 80 kHz or higher.
I decided to take a moment to consider this.
It's been remarked on correctly that humans cannot hear above 20 kHz as a theoretical limit, and that for the vast majority of humans, the actual limit is certainly quite a bit lower than that. I'll also note that 80 kHz is in fact well above the theoretical limit of hearing
for a dog.
I've had some experience in the music world as a professional composer and musician, and I've dabbled in recording engineering myself. I've recorded professionally as classical performer as well, and I've very rarely seen in actual use microphones that are sensitive enough to pic up frequencies above 20 kHz at any decent dB level. A quick perusal of any of the most commonly preferred microphones used by the top recording engineers all have documented steep fall offs in terms high frequency response, usually well before 20 kHz. For example the famous Neumann U87s are at -10 db already at 20 kHz, and falling fast. Those are considered "vintage," but with tons of clones that are now commonly in use, included updated versions from Neumann, such as their U89: these all fall off significantly well before 20 kHz.
There are some mics available, such as the Sennheiser MKH 800, that actually do capture information at a reasonable dB above 20 kHz, but to my knowledge and experience they are actually not at all very common as a preference in the industry, oddly enough not even for high-resolution formats. Certainly there's no guarantee such mics are included in those recordings; when recordings have included mic information, the consistently most usual choices in my experience are not these ultra high-resolution mics. Maybe someone in the industry here can chime in as to why this is?
In any case, certainly Red Book CDs cannot include any sounds above 22 kHz. That's a hard cut-off, a consequence of the Nyquist Theorem. In fact, any frequencies in the signal above 22 kHz will "fold over" at equal amplitude at the difference between the signal and 22 kHz, subtracted from 22 kHz. Frequencies above 22 kHz are definitely therefore something the engineer would want to avoid, with an eye towards adhering to the Red Book standard, at some point in the mastering process.
Higher-resolution formats such as SACD or Blu-ray Audio can certainly include information above 22 kHz, if the source recording was made with microphones that have that sensitivity. Again, this is not at all guaranteed to be the case.
And let's be clear: the sound energy produced by ordinary instruments' harmonic spectra at such frequencies is
extremely low! ETA: likely well below an audible range in dB terms even if human hearing included those frequencies.
Vinyl as a playback medium certainly can't come close to 22 kHz, and to my knowledge there are no purely
acoustic recording formats that can.
And even with high-resolution formats, as far as I can tell there is not actually sufficiently consistent usage of those ultra high-frequency mics to be sure to be sure these recordings include information above 22 kHz (I could be wrong about this but it sure doesn't look like it). And they certainly won't when "mixed down" to the Red Book standard, nor indeed if they are sourced from an originally acoustic master.
What about speakers? Perusing the specs of most of even the most high-end speakers and monitors, I noticed that their frequency response almost always falls off of a cliff at 20 kHz. True, there are some studio monitors such as the Focal Solo6 BEs that are rated at -3dB at 40 kHz. But the overwhelming majority of speakers and monitors in people's homes and recording studios include tweeters that definitely fall off very quickly indeed above 20 kHz.
Those hounding for ultra high-frequency bandwidth measurements of amplifiers are therefore crying out for measuring distortion at ultra high frequencies at dB levels that:
- definitely all humans (and probably most dogs) cannot hear.
- most microphones cannot pick up.
- most playback formats cannot include.
- the overwhelming majority of tweeters cannot reproduce.
The first point explains the remainder. It doesn't make much sense to engineer for frequencies no one can hear; that's ultimately just a waste of effort and resources.*
I think it's fair to be persuaded by the measurement standards Amir is employing in his reviews.
Furthermore, I see zero reason to slander amplifier manufacturers such as Hypex and Purifi with thinly-veiled accusations of fraud,
especially from someone who is also trying to sell their own brand of amplifiers.
I conclusion, A7 of this FAQ certainly answers this question to my satisfaction.
(*However, I'll concede that there do appear to be some advantages in the mastering stages of a digital recording to make use of higher bit depth than the Red Book standard 16.)