HiRes FOMO - Am I really missing anything?

[OldTimer]

OK. I gave CD as an example. FLAC and ALAC are losslessly compressed so you have the same audio content in ~60% of the data, which would be 800-900 kb/s (it varies with the music content). To get the bitrate you need only to multiply sample rate x bit depth x channels, so for 96 kHz, 24 bit, 2 channels it would be 4.6 Mb/s, which I would expect to compress to ~2.8 Mb/s FLAC, and for 192 kHz simply double those.

The issue is whether this is literally a waste of space. Is there a provable audible difference compared with CD quality, with everything else the same, of course?

Of course, the bitrate. How many bits will be fed to the system in every second. DACs are hungry of bitrate. Amps are hungry of power. And so on.

 

[welwynnick]

Hypersonic refers to very high speed, so ultrasonic is a better expression.

This test by Googlebot on Hydrogen Audio (where ABX is a way of life) is interesting, not just for the multiple statistically significant results.
The OP was able to hear the difference between 96k and 44.1k sampling,
He could only do this using his Elac (50kHz) speakers, and not his Canton (20kHz) speakers. Fair enough.
However, he admitted himself that his hearing topped out at 17kHz.
Therefore even 20kHz was ultrasonic for him, and the speakers shouldn't have made any difference.

 

[vintologi]

Hypersonic refers to very high speed, so ultrasonic is a better expression.

This test by Googlebot on Hydrogen Audio (where ABX is a way of life) is interesting, not just for the multiple statistically significant results.
The OP was able to hear the difference between 96k and 44.1k sampling,
He could only do this using his Elac (50kHz) speakers, and not his Canton (20kHz) speakers. Fair enough.
However, he admitted himself that his hearing topped out at 17kHz.
Therefore even 20kHz was ultrasonic for him, and the speakers shouldn't have made any difference.

Does the elac speakers have the same or lower additional IMD from the ultrasonics than the canton ones?

Edit: googlebot unfortunately also reduced the bit-rate to 16 which could have influenced the result as well.

 

[welwynnick]

Does the elac speakers have the same or lower additional IMD from the ultrasonics than the canton ones?

If IMD was the explanation for the audible difference ( by adding distortion harmonics into the audible band ) then the Canton speakers would have reproduced it.
It's not impossible that the Elac were distorting more than the Canton, but it's also possible that it's the other way round - in fact I think it's more likely.
All HiFi equipment generally performs well in the middle of it's audio envelope. As you approach the limits of amplitude and low and high frequency, the distortion generally goes up. It's not a cast iron rule, but exceptions are like rocking horse teeth. Since the Cantons are rated to 20k and the Elacs to 50k, when playing hi-res audio, it's fair to say that since the Elacs are working within their operating bandwidth, and the Canton would be trying to operate outside, that it's the Cantons and not the Elacs that are more likely to generate excess IMD. Therefore if IMD is the explanation for the audible difference, then that would have been audible with the Cantons.
The reason that I flagged up Googlebot's test is because he DIDN'T hear the difference with the 20kHz speakers.
Hydrogen Audio is full of ABX test results, and there are many more like this.

 

[Redacted]

Does anyone listen to music?

 

[vintologi]

One issue with using a separate tweeter for the ultrasonics is that it will cause interference issues at the crossover area and you end up having to have your head in an exact position for it to work well.

Another approach is to have one speaker always play the the track with the ultrasonics intact and have the other speaker play just the ultrasonics at inverted phase.

Let me know if you want me to make a test-track for some specific song for you.

 

[vintologi]

If IMD was the explanation for the audible difference ( by adding distortion harmonics into the audible band ) then the Canton speakers would have reproduced it.
It's not impossible that the Elac were distorting more than the Canton, but it's also possible that it's the other way round - in fact I think it's more likely.
All HiFi equipment generally performs well in the middle of it's audio envelope. As you approach the limits of amplitude and low and high frequency, the distortion generally goes up. It's not a cast iron rule, but exceptions are like rocking horse teeth. Since the Cantons are rated to 20k and the Elacs to 50k, when playing hi-res audio, it's fair to say that since the Elacs are working within their operating bandwidth, and the Canton would be trying to operate outside, that it's the Cantons and not the Elacs that are more likely to generate excess IMD. Therefore if IMD is the explanation for the audible difference, then that would have been audible with the Cantons.
The reason that I flagged up Googlebot's test is because he DIDN'T hear the difference with the 20kHz speakers.
Hydrogen Audio is full of ABX test results, and there are many more like this.

Honestly i think that ABX result is stronger evidence than any formal study i have seen so far.

Makes it worthwhile for me to see if i can get a similar result but fully controlled for IMD. I might however need some more training for this (i also plan on getting an amplifier with better ultrasonic response eventually).

Since we suspect that the difference is due to 24-96 offering better transient response i will change methodology to playing a mono version of the track in one speaker (always full resolution) and then using the other speaker to sabotage it by playing the ultrasonics out of phase for one version and have the speaker be silent in the other.

 

[welwynnick]

One issue with using a separate tweeter for the ultrasonics is that it will cause interference issues at the crossover area and you end up having to have your head in an exact position for it to work well.

I believe the cross-over to the tweeter is at a fairly normal 2.3 or 2.7 kHz (sources vary).

 

[vintologi]

I believe the cross-over to the tweeter is at a fairly normal 2.3 or 2.7 kHz (sources vary).

I am referring to the use of a separate tweeter just for the ultrasonics in order to control for IMD.

My old speakers (dynavoice DM5) had ribbon tweeters crossed at 8Khz. My current tweeters (TW29BN) should have lower distortion at a given frequency but they also do have to work a lot harder at the lower frequencies (they are crossed a lot lower and i think the XO is less steep as well).

I believe elac speakers with AMTs are crossed around 3Khz.

Edit: here is a link to the files where one has a sabotage channel containing ultrasonics that are the difference from resampling inverted:

932.87 MB folder on MEGA

3 files

mega.nz

Full album can be downloaded here:

Suitzer Harp, by GKMHarp

16 track album

gkmharp.bandcamp.com

 

[RandomEar]

Since the Cantons are rated to 20k and the Elacs to 50k, when playing hi-res audio, it's fair to say that since the Elacs are working within their operating bandwidth, and the Canton would be trying to operate outside, that it's the Cantons and not the Elacs that are more likely to generate excess IMD.

I don't think that's correct to assume (but I'm happy to be corrected in case it is): If the tweeter can't reproduce the frequency, why would that cause IMD? The tweeter is either low-passed in the crossover or acts as a mechanical low pass filter itself here. And frequencies which dont reach it or which it can't reproduce also can't interact with those which are reproduced. They are simply dissipated as heat in the crossover or the voice coil. So no IMD, because tones which don't exist can't intermodulare with others, right?

On the contrary: If the ELACs can reproduce 30 or 40 kHz, those tones can produce IMD because they are being played at relevant levels. I don't think this matters for the test, though. There are differences between the speakers, so it's not really important which one potentially has IMD due to ultrasonic noise.

---

Different topic: The "transient response" argument is always weird to me from a logical perspective. We can't hear past 20-ish kHz, because our hearing and in particular most likely the hair cells are not sensitive to those frequencies. They act akin to a mechanical low pass. So, in consequence, the sharpest transient we will ever be able to hear is the gradient close to the zero crossing of a 20-ish kHz sine wave. You cannot logically separate "transient response" from "frequency response". They are inherently the same thing.

 

[101geo]

Does anyone listen to music?

I do and I prefer 24/96. I "know" it's the same but I prefer it. I know I shouldn't and refuse to do a blind test but I'm old, retired and spoiling myself now.

 

[Phorize]

I can't hear any obvious differences between a Spotify stream in 320kbps ogg vorbis and 44.1/24 flac on my admittedly not full range system in my very compromised listening space. The result is that have lost all interest in the subject for purposes of practical audio application. Obviously it's an interesting area scientifically.

 

[welwynnick]

I don't think that's correct to assume (but I'm happy to be corrected in case it is): If the tweeter can't reproduce the frequency, why would that cause IMD? The tweeter is either low-passed in the crossover or acts as a mechanical low pass filter itself here. And frequencies which dont reach it or which it can't reproduce also can't interact with those which are reproduced. They are simply dissipated as heat in the crossover or the voice coil. So no IMD, because tones which don't exist can't intermodulate with others, right?

Tweeters are usually high-passed in the crossover so they don't get over-loaded with too much LF excursion.
A big bass driver HF response tends to be limited by it's inductance rather than cone break-up, as the wavelength of sound is much bigger than the driver.
Mid-range and tweeter drive unit dimensions are generally similar to the shortest wavelengths they reproduce.
Their HF limit is generally set by cone or dome break up, and that definitely amounts to distortion.
It's not impossible that a tweeter, being pushed beyond it's dome breakup frequency, has less distortion than a ribbon tweeter working inside it's working range, but I think the likelihood of that is vanishingly small.

Different topic: The "transient response" argument is always weird to me from a logical perspective. We can't hear past 20-ish kHz, because our hearing and in particular most likely the hair cells are not sensitive to those frequencies. They act akin to a mechanical low pass. So, in consequence, the sharpest transient we will ever be able to hear is the gradient close to the zero crossing of a 20-ish kHz sine wave. You cannot logically separate "transient response" from "frequency response". They are inherently the same thing.

Yes, transient response and frequency response are convolutions of each other, but only when the phase response is linear.
If the frequency response is the same and the phase response is different, then the transient response will be different.
Of course this might be what's happening with Googlebot's test - it could be that the DAC output or the speaker response are linear in phase, and that's what's heard.
When you band limit the frequency response of anything, you always introduce a phase shift as well. That applies to both the limit of LF response, and to the limit of HF response. If either are extended, even to frequencies where you can't hear the sine wave (like Googlebot can't hear over 17kHz) you will still get a better phase response.

 

[voodooless]

So, that IMD argument... how would this actually work? Because all that ultrasonic noise is basically down to like 70 to 90 dBFS.. This is way below the noise floor of most rooms at normal listening levels already. Any IMD would have to be of an even lower level... Never mind that 30 kHz sound already loses 1 dB per meter due to friction (atmospheric absorption). It gets worse if the frequency goes up.

 

[Redacted]

I do and I prefer 24/96. I "know" it's the same but I prefer it. I know I shouldn't and refuse to do a blind test but I'm old, retired and spoiling myself now.

It's only the "same" if it's from the same stereo master as say a 16/44 might be - some 24/96's may be from different masters, as widely discussed. Also, I've found that if I buy certain FLACs instead of a CD, I can't even buy a 16/44 version and only 24/44 or 24/96 are available - I too am past caring so long as I like the album! I'm just not going down the rathole of DSD or MQA (or SACD) or whatever the latest fashion format is!
Sorry - I'm still off-topic!

 

[vintologi]

So, that IMD argument... how would this actually work? Because all that ultrasonic noise is basically down to like 70 to 90 dBFS.. This is way below the noise floor of most rooms at normal listening levels already. Any IMD would have to be of an even lower level... Never mind that 30 kHz sound already loses 1 dB per meter due to friction (atmospheric absorption). It gets worse if the frequency goes up.

There can be significant ultrasonics (depending on the track) which may produce IMD that would be audible by itself (even if it's very difficult to pick up when it's mixed with the rest of the music. Generally the more powerful the ultrasonics are the more likely they are to create an audible difference due to IMD produced somewhere (DAC, amplifier, tweeter, air, the ear, etc).

IMD emerging from the speaker of before that is unwanted since that distorts the sound compared to if you had heard it live.

As far as i know there is no evidence for people being able to hear a difference from ultrasonics due to non-linearities in the air, reflections or the ear (perhaps some people can but it's obviously going to be very hard to do).

 

[AudioSceptic]

Does anyone listen to music?

How dare you ask such an outrageous question! ;-)

 

[voodooless]

As far as i know there is no evidence for people being able to hear a difference from ultrasonics due to non-linearities in the air, reflections or the ear (perhaps some people can but it's obviously going to be very hard to do).

Sure there is evidence! There are even devices that specifically work on this principle, like this one:

True directional speakers using modulated ultrasound | MadPsy's Place

Now, if this would be possible to do with a normal speaker, is a totally different matter. And yes, any of these effects would be unwanted, no matter if they are subjectively better or not.

 

[welwynnick]

This test by Googlebot on Hydrogen Audio (where ABX is a way of life) is interesting, not just for the multiple statistically significant results.

Does the elac speakers have the same or lower additional IMD from the ultrasonics than the canton ones?

One other point to make about Googlebot's test is that two other people got statistically significant results (p< 0.05) when using completely different systems.

It's not as if it's an isolated test, either. I found a dozen similar tests on HA before I stopped looking

 

[Redacted]

How dare you ask such an outrageous question! ;-)

Owing to FOMO, I spend hours every day, straining to hear like a bat can, or setting up my earthquake sensors, or literally counting the bits passing on my ethernet cable

I have enough evidence from messing with my friends (with audio interests) that most of them can't tell a 320kbps MP3 from a 24/192 FLAC - but then we only listen to pink noise (weighted) that was produced between 1970 and 1990