Is Hi-Res Audio better because filter is outside audible range?

[danadam]

There are ways to not use a low-pass filter when downsampling an already digitized audio file, which can cause aliasing.

I was talking in the context of actual devices, but maybe I shouldn't have narrowed it so much. I know that upsampling can create images and downsampling can create aliases .

 

[bhobba]

Not because it is outside the audible range (even 44.1/16 contains the audible range - except maybe for dogs) as much as it is at a frequency where everything is noise that swamps actual audio. Several DACs, some generally considered among the world's best (eg the Chord Dave), have a filter at 50 kHz. This is because everything above is noise, except for a rare, exceptional recording. If you use such a DAC, over 96k sampling is useless.

DXD has been found to produce very high-quality recordings:

Lower Resolution, Better Sound Quality?

TRPTK • Recording the extraordinary. High-resolution, immersive recordings of contemporary artists in classical music and jazz. Buy hi-res downloads up to DXD 32bit, Dolby Atmos TrueHD, DTS:X, Auro-3D, Hybrid SACDs, and vinyl.

trptk.com

The trouble is that they are mainly noise that does not compress well using commonly available compression methods such as FLAC.

What About DXD? Surprise! – Real HD-Audio

www.realhd-audio.com

However, there is a compression method that can transmit remove a lot of the noise and not affect the audio. It is lossyWAVE followed by FLAC:

https://wiki.hydrogenaud.io/index.php?title=LossyWAV

 

[welwynnick]

I've been meaning to reply to this thread for some time, not so much for the debate as to get something off my chest.

We're saddled with CD quality audio because that was the state of the art during the 1980's.

Across recording, mastering, distribution, storage and playback, we had better formats very soon afterwards, but Sony and Philips were impatient.

CD bandwidth extends to 20kHz, and I certainly can't hear above 20kHz any more, but is that all that counts?

I've been struggling to reconcile subjective and objective measures for a long time, but think I'm getting somewhere.

I now think measurements matter more than I ever thought before, but they have to be the right measurements.

It's convenient to focus on the most favourable measures at 1V 1W 1% 1dB 1kΩ 1kHz 1µV etc. but I think we need to look harder.

I think what really counts is how fidelity is maintained around the whole of the audio envelope - a virtual cube that characterises amplitude, frequency and load.

And fidelity has several essential parameters - amplitude linearity, frequency linearity, and phase linearity.

My view is that good audio requires an envelope that encompasses everything we can hear, plus linear responses over the whole of that envelope (not just in the middle).

There's traditionally been more focus on amplitude and frequency linearity, but I think phase linearity is important too.

Anyone who's heard a good infinite baffle subwoofer will know how linear phase response and low group delay makes bass sound better.

In the mid range, speaker cross-overs often mess up phase response where our ears are most receptive, and single drive unit speakers like electrostatics have magical transparency in the mid range (if less so at frequency extremes). However, really effective DSP like Trinnov, Audiolense, Dirac, Neumann, Genelec etc can help to restore that.

At high frequencies, format limitations come into play, because there isn't enough room for the reconstruction filter between 20kHz and the Nyquist frequency, 22.05kHz.

We want amplitude, frequency and phase linearity at all frequencies, but DAC filters can only get two out of three right.

There are lots of options - linear phase, minimum phase, fast roll-off, slow-roll-off, hybrids etc. Each one is a compromise, and at 44.1kHz you never get all 3 linearities right.

The reason for this post is because here at ASR, jkim has posted his detailed measurements of the high performing JCALLY JM20 MAX: USB-C Headphone Dongle.

Like some of the recent Soundstage reviews, he measured amplitude AND phase response vs frequency.

With 44.1kHz sampling, the FR is flat to 20kHz, but the phase response deviates by hundreds of degrees within the audio bandwidth.

With 96kHz sampling, the FR is still flat to 20kHz and beyond, but the phase only deviates by tens of degrees within the audio bandwidth.

Soundstage reviews show a wide range of compromised results.

Why does this matter? It doesn't matter for frequency response and tonal accuracy, but it directly impacts transient and time domain behaviour.

While it's true that uniform frequency response is essential for accurate transient response, uniform phase response is essential, too.

So, no, it's not about ultrasonic hearing and golden ears and all that, its about get getting all aspects of fidelity right WITHIN the audio envelope.

 

[DVDdoug]

While it's true that uniform frequency response is essential for accurate transient response, uniform phase response is essential, too.

I don't believe so. Experiments have been done with all-pass filters that shift-rotate phase without affecting the frequency response. It doesn't change the sound.

Phase is only a problem when there are two sounds at the same frequency at different phases. An extreme case is if you flip the phase-polarity of one speaker the out-of-phase soundwaves cancel and interfere giving a weird "spacey" sound, and the bass will cancel almost completely. If you flip the polarity of both speakers everything sounds OK again.

 

[antcollinet]

Hi-res audio sounds better because of its ultrasonic content

Before asking "why" it sounds better, you really need to check "if" it sounds better.

All the evidence suggests there is little to no difference to hear between redbook 44.1/16 and higher res formats.

My expectation if there is a real difference, it will most likely come from mastering - as suggested by this article from @amirm.

High Resolution Audio: Does It Matter?

High Resolution Audio: Does It Matter? By Amir Majidimehr [Note: This article was published in the January 2015 issue of Widescreen Review Magazine. This is a slightly revised edition.] Yes! Thought I give you the short answer first in case you are impatient like me. If you have read my...

www.audiosciencereview.com

Certainly any real differences stemming purely from the format are going to marginal at best.


See also:

24/192 Music Downloads are Very Silly Indeed

people.xiph.org

 

[danadam]

There are lots of options - linear phase, minimum phase, fast roll-off, slow-roll-off, hybrids etc. Each one is a compromise, and at 44.1kHz you never get all 3 linearities right.

What is not right with linear phase filter?

 

[Keith_W]

I don't believe so. Experiments have been done with all-pass filters that shift-rotate phase without affecting the frequency response. It doesn't change the sound.

Phase is only a problem when there are two sounds at the same frequency at different phases. An extreme case is if you flip the phase-polarity of one speaker the out-of-phase soundwaves cancel and interfere giving a weird "spacey" sound, and the bass will cancel almost completely. If you flip the polarity of both speakers everything sounds OK again.

Sorry Doug this is incorrect. Please check out JJ's comments in the latter part of this thread and what he says here. He has said somewhere else that phase deviation of > 15 degrees per ERB is audible. In short, both intrachannel and interchannel phase anomalies are audible.

 

[MRC01]

...
There's traditionally been more focus on amplitude and frequency linearity, but I think phase linearity is important too.
...
There are lots of options - linear phase, minimum phase, fast roll-off, slow-roll-off, hybrids etc. Each one is a compromise, and at 44.1kHz you never get all 3 linearities right.
...

The phase response of DACs is easy to measure and with a linear phase filter the response is usually dead-flat in the audible band. Some DACs have a minimum phase filter which is not flat, but that is a preference, not a technology limitation.

For example, here's the phase response of one my DACs, which uses the outdated Wolfson WM8741. The top line is the frequency response (left Y axis), the bottom line is phase (right Y axis). This uses the DAC's built-in linear phase filter. I eyeball that as roughly +/- 3* around zero, near perfect.

This DAC chip has 5 different built-in filters. We can switch it to use a different one that is minimum phase. Then we get the second graph below: much bigger phase shifts. But this is an optional choice, not a technology limitation.

The real limitation of 44.1 kHz sampling is that the filter transition band is so narrow that a proper filter is hard to implement if it has to run in real-time on a DAC chip. That's why most of the DACs we see (including the WM8741 shown above) don't fully attenuate by 22,050 Hz (Nyquist) like they should. Instead they stretch the transition band beyond Nyquist to 24,100 Hz (or so). This can leak ultrasonic noise and trigger IM distortion in the audible passband. From an engineering perspective we can call this an imperfect filter, but whether it causes any audible problems is debatable.

Either way, it doesn't take much to get a truly proper filter. Sample slightly faster at 48 kHz to get a slightly wider transition band and the problem disappears. Most of the filters on these DACs (including the WM8741 shown above) fully attenuate by Nyquist at 48 kHz and higher rates.

I'd like to see 48 kHz become the standard for audio. We'd have proper filters in our DACs (whether or not the improvement would be audible) and the simplicity of having the same format for CD, DVD and streaming.

 

[Audiofire]

I'd like to see 48 kHz become the standard for audio.

I don't know what is meant by standard here. Many downloadable releases use the given sample rate. I do like this sample rate too.

 

[danadam]

This can leak ultrasonic noise and trigger IM distortion in the audible passband. Sample slightly faster at 48 kHz to get a slightly wider transition band and the problem disappears.

As far as the danger of IM distortion is concerned, is there a difference between ultrasonics from leaked images of 44.1k sampling rate and in-band ultrasonics of 48k samping rate?

Most of the filters on these DACs (including the WM8741 shown above) fully attenuate by Nyquist at 48 kHz and higher rates.

Not in my experience (but I mostly have cheap dongles). Here's 1 minute of white noise at 48k sampling rate and -8 dBFS peak, played by several DACs and recorded by ADCiso at 96k sampling rate:

In my case only JCally JM6, Apple and Samsung dongles attenuate before 24k. For completeness here's when they play noise at 44.1k sampling rate (but not Apple dongles because they don't support anything other than 48k):

BTW, the poor results of Apple dongles are because of something that on a spectrogram looks like clipping. I tried with much lower level but it didn't help. Here's comparison of the first 10 seconds between Apple dongle US and Samsung dongle:

 

[MRC01]

I don't know what is meant by standard here. Many downloadable releases use the given sample rate. I do like this sample rate too.

By "standard" I mean 44.1 kHz is abandoned and replaced by 48 kHz. For example CDs become 48 kHz too.
I know, it's just a pipe dream.

... Most of the filters on these DACs (including the WM8741 shown above) fully attenuate by Nyquist at 48 kHz and higher rates.

... Not in my experience (but I mostly have cheap dongles). Here's 1 minute of white noise at 48k sampling rate played by several DACs and recorded by ADCiso at 96k sampling rate:

In addition to my old WM8741 DAC I've tested a few of the better DACs that AMIR has reviewed here like the Topping E70, EX5 and SMSL SU-6. At 44.1 kHz they stretch the filter transition band past Nyquist, but at 48 kHz they fully attenuate by Nyquist. But of course this will vary depending on the chip and the DAC.

 

[MRC01]

As far as the danger of IM distortion is concerned, is there a difference between ultrasonics from leaked images of 44.1k sampling rate and in-band ultrasonics of 48k samping rate?

Yes, I think. In one case the DAC can produce tones that are not encoded in the recording. In the other, the DAC correctly produces what is in the recording. In this second case, if the recording is clean then playback will be too, and whether there are ultrasonics is up to the recording producers/creators.

 

[Keith_W]

BTW all those people who think that hi-res audio is better should read this article. The first half is a fun read, but the eye opening bit starts at the section "192kHz considered harmful". Quote from that article:

192kHz digital music files offer no benefits. They're not quite neutral either; practical fidelity is slightly worse. The ultrasonics are a liability during playback.

Neither audio transducers nor power amplifiers are free of distortion, and distortion tends to increase rapidly at the lowest and highest frequencies. If the same transducer reproduces ultrasonics along with audible content, any nonlinearity will shift some of the ultrasonic content down into the audible range as an uncontrolled spray of intermodulation distortion products covering the entire audible spectrum. Nonlinearity in a power amplifier will produce the same effect. The effect is very slight, but listening tests have confirmed that both effects can be audible.

Above: Illustration of distortion products resulting from intermodulation of a 30kHz and a 33kHz tone in a theoretical amplifier with a nonvarying total harmonic distortion (THD) of about .09%. Distortion products appear throughout the spectrum, including at frequencies lower than either tone.
Inaudible ultrasonics contribute to intermodulation distortion in the audible range (light blue area). Systems not designed to reproduce ultrasonics typically have much higher levels of distortion above 20kHz, further contributing to intermodulation. Widening a design's frequency range to account for ultrasonics requires compromises that decrease noise and distortion performance within the audible spectrum. Either way, unneccessary reproduction of ultrasonic content diminishes performance.
There are a few ways to avoid the extra distortion:

1. A dedicated ultrasonic-only speaker, amplifier, and crossover stage to separate and independently reproduce the ultrasonics you can't hear, just so they don't mess up the sounds you can.
2. Amplifiers and transducers designed for wider frequency reproduction, so ultrasonics don't cause audible intermodulation. Given equal expense and complexity, this additional frequency range must come at the cost of some performance reduction in the audible portion of the spectrum.
3. Speakers and amplifiers carefully designed not to reproduce ultrasonics anyway.
4. Not encoding such a wide frequency range to begin with. You can't and won't have ultrasonic intermodulation distortion in the audible band if there's no ultrasonic content.

They all amount to the same thing, but only 4) makes any sense.

 

[MRC01]

BTW all those people who think that hi-res audio is better should read this article. The first half is a fun read, but the eye opening bit starts at the section "192kHz considered harmful". Quote from that article:

I've seen this happen in my own testing. A certain device showed massive HF noise but only at 176.4 and 192 kHz sampling rates. Long story short, it turns out that it had a switching power supply with a frequency of 65 kHz that was leaking through into the audio. At sample rates of 96k and below it measured clean because that noise was filtered out.

That said, the definition of "high res" is usually anything more than 16-bit 44.1 kHz sampling. I do think 24-bit 48 kHz would be an improvement. But we don't need to get crazy about it.

 

[manisandher]

We're saddled with CD quality audio because that was the state of the art during the 1980's.

CD quality is more than good enough for replay in the home. IMHO and experience.

The recording quality and mastering ulitmately determine the SQ.

 

[welwynnick]

CD quality is more than good enough for replay in the home. IMHO and experience.
The recording quality and mastering ulitmately determine the SQ.

Have you ever heard blu-ray audio through electrostatic speakers?

 

[antcollinet]

Have you ever heard blu-ray audio through electrostatic speakers?

Stereo? It'll sound just the same as CD through electrostatic speakers - if the mastering is the same.

 

[oleg87]

Have you ever heard blu-ray audio through electrostatic speakers?

My hearing taps out around 15khz these days (and if the age in your profile is right, I doubt your ears would fare better), so a medium's ability to reproduce near-ultrasound is going to be lost on me (and I have done blind tests confirming as much).

(incidentally even 8 bit depth is adequate for my ears these days, since at 44.1khz you can noise shape the dither mostly out of my hearing range - wasn't the case when I was messing around with audio in my teens)

 

[welwynnick]

Stereo? It'll sound just the same as CD through electrostatic speakers - if the mastering is the same.

So you haven't heard it then?
Please spare me the inevitable reply, I know exactly what it will be.

 

[welwynnick]

My hearing taps out around 15khz these days (and if the age in your profile is right, I doubt your ears would fare better), so a medium's ability to reproduce near-ultrasound is going to be lost on me (and I have done blind tests confirming as much).

Me too.
It's ironic that the only people that care about high fidelity are the one who won't benefit.