High resolution audio. The science, or lack of...?

[davedotco]

In order to move this dicussion onwards we need to find a way to look at this issue in a sensible and productive manner.

This thread has thrown up the following 'facts'.

Nyquist-Shannon states quite clearly that Red Book standard processing will recreate the original waveform exactly, I am comfortable with that.

It is also clear that certain other processes need to be carried out, I am thinking of anti-aliasing filters and noise shaping primarily but there are others.

These processes have a measureable effect and increase noise and distortion. A greater bit rate and sampling frequency will reduce these effects.

Therefore it is clear that 24/96 playback will be superior to 16/44.1 in terms of measurement. Again I have no problem with that.

Many people (myself included) have tried various tests and found it very dificult to distinquish 24/96 from 16/44.1 all other factors being equal.

Put simply, noise and distortion is present in greater quantity in 16/44.1 than it is in 24/96 but is inaudible to many listeners.

So I was looking at a way that this difference could be correlated and perhaps measured. I have heard the phrase 'effectively transparent' used in similar contexts.

So if we were to measure distortion and noise, is there a figure that could be agreed on, below which these distortions are inaudible and the process (or component) deemed effectively transparent?

 

[fr0g]

In order to move this dicussion onwards we need to find a way to look at this issue in a sensible and productive manner.

This thread has thrown up the following 'facts'.

Nyquist-Shannon states quite clearly that Red Book standard processing will recreate the original waveform exactly, I am comfortable with that.

It is also clear that certain other processes need to be carried out, I am thinking of anti-aliasing filters and noise shaping primarily but there are others.

These processes have a measureable effect and increase noise and distortion. A greater bit rate and sampling frequency will reduce these effects.

Therefore it is clear that 24/96 playback will be superior to 16/44.1 in terms of measurement. Again I have no problem with that.

Many people (myself included) have tried various tests and found it very dificult to distinquish 24/96 from 16/44.1 all other factors being equal.

Put simply, noise and distortion is present in greater quantity in 16/44.1 than it is in 24/96 but is inaudible to many listeners.

So I was looking at a way that this difference could be correlated and perhaps measured. I have heard the phrase 'effectively transparent' used in similar contexts.

So if we were to measure distortion and noise, is there a figure that could be agreed on, below which these distortions are inaudible and the process (or component) deemed effectively transparent?

Nice summing up.

Pretty much in 100% agreement here.

What we could also do is take the 24/96 track and the 16/44.1 conversion and extract the difference as a WAV file. Play that back...

Do you hear anything?

If you hear something here, then it is "possible" you will hear a difference between the tracks. If you don't, then it is quite clearly an inaudible difference.

Something I might try at the weekend.

 

[davedotco]

In order to move this dicussion onwards we need to find a way to look at this issue in a sensible and productive manner.

This thread has thrown up the following 'facts'.

Nyquist-Shannon states quite clearly that Red Book standard processing will recreate the original waveform exactly, I am comfortable with that.

It is also clear that certain other processes need to be carried out, I am thinking of anti-aliasing filters and noise shaping primarily but there are others.

These processes have a measureable effect and increase noise and distortion. A greater bit rate and sampling frequency will reduce these effects.

Therefore it is clear that 24/96 playback will be superior to 16/44.1 in terms of measurement. Again I have no problem with that.

Many people (myself included) have tried various tests and found it very dificult to distinquish 24/96 from 16/44.1 all other factors being equal.

Put simply, noise and distortion is present in greater quantity in 16/44.1 than it is in 24/96 but is inaudible to many listeners.

So I was looking at a way that this difference could be correlated and perhaps measured. I have heard the phrase 'effectively transparent' used in similar contexts.

So if we were to measure distortion and noise, is there a figure that could be agreed on, below which these distortions are inaudible and the process (or component) deemed effectively transparent?

Nice summing up.

Pretty much in 100% agreement here.

What we could also do is take the 24/96 track and the 16/44.1 conversion and extract the difference as a WAV file. Play that back...

Do you hear anything?

If you hear something here, then it is "possible" you will hear a difference between the tracks. If you don't, then it is quite clearly an inaudible difference.

Something I might try at the weekend.

Nice idea, way beyond my compedence though.

If the 'difference' is measureable then we might be able to put a figure on the level needed to make an audible difference. Interesting.

I was lucky enough, back in the 70s, to see Bob Carver perform his 'nulling' technique when demonstrating his famour 'amplifier matching' challenge that was written up in Stereophile.

 

[jcbrum]

What we could also do is take the 24/96 track and the 16/44.1 conversion and extract the difference as a WAV file. Play that back...

Do you hear anything?

If you hear something here, then it is "possible" you will hear a difference between the tracks. If you don't, then it is quite clearly an inaudible difference.

Something I might try at the weekend.

Hmmm, not sure that is valid. You would have to be able to audibly detect the 'difference' signal whilst the 16bit file is playing at the same time, to reflect the real conditions.

For example, Oldrics distortion theory suggests (and his post illustrates) that distortion is at -90dB. That is even totally inaudible on it's own, let alone when full (normal) volume music sound is being played back.

Someone called this a 'slam-dunk', but of course it is far from being so. -90dB is irrelevant to human ears in domestic environments.

JC

 

[fr0g]

What we could also do is take the 24/96 track and the 16/44.1 conversion and extract the difference as a WAV file. Play that back...

Do you hear anything?

If you hear something here, then it is "possible" you will hear a difference between the tracks. If you don't, then it is quite clearly an inaudible difference.

Something I might try at the weekend.

Hmmm, not sure that is valid. You would have to be able to audibly detect the 'difference' signal whilst the 16bit file is playing at the same time, to reflect the real conditions.

For example, Oldrics distortion theory suggests (and his post illustrates) that distortion is at -90dB. That is even totally inaudible on it's own, let alone when full (normal) volume music sound is being played back.

Someone called this a 'slam-dunk', but of course it is far from being so. -90dB is irrelevant to human ears in domestic environments.

JC

Thus "possible". It's an interesting test anyway to see just how audible the difference, played back in a silent room, actually is.

This is a test on the difference between 2 lossless formats...not MP3, in which I think your point would be 100% valid. However we are saying (I think) that the distortion is in the 20 KHz plus region, and very quiet...There isn't a cat in hell's chance I will hear it.

 

[jcbrum]

Hmmm, MP3 compression is an entirely different concept to 16bit v 24bit, save for the possibility of loud sounds masking quiet ones.

Some readers might have difficulty understanding how quiet -90dB actually is.

The difference between microphone or cartridge level and line level is around 40dB.

The difference between line level and speaker level is also around 40dB.

Added together that makes 80dB.

A signal at -90dB would be worse (quieter) than connecting your microphone or cartridge directly to your loudspeakers, in your normal listening environment, without any amplification at all.

JC

 

[fr0g]

Hmmm, MP3 compression is an entirely different concept to 16bit v 24bit, save for the possibility of loud sounds masking quiet ones.

Some readers might have difficulty understanding how quiet -90dB actually is.

The difference between microphone or cartridge level and line level is around 40dB.

The difference between line level and speaker level is also around 40dB.

Added together that makes 80dB.

A signal at -90dB would be worse (quieter) than connecting your microphone or cartridge directly to your loudspeakers, in your normal listening environment, without any amplification at all.

JC

Yep,. I know, and why I said the test would be invalid. I have seen an advert for a lossless streaming service that does exactly what I suggested earlier to advertise how much better lossless is, when in actual fact, those sounds would be inaudible in real life.

 

[fr0g]

As an aside, I am currently contemplating swapping out my Squeezebox for one of either Sonos or Bluesound.

One negative about the Sonos is that it won't play HD music...

I do own some and would still buy it if I knew the mastering was better...

Just a shame the Bluesound has no Spotify client or it would be a shoe-in.

 

[John Duncan]

As an aside, I am currently contemplating swapping out my Squeezebox for one of either Sonos or Bluesound.

One negative about the Sonos is that it won't play HD music...

I don't know anything about Bluesound, but the Sonos is so mature in terms of UX I can highly recommend it. I don't use music above 320k on it, but I only have a Play:3 (for now - I can feel a Connect coming on).

 

[fr0g]

As an aside, I am currently contemplating swapping out my Squeezebox for one of either Sonos or Bluesound.

One negative about the Sonos is that it won't play HD music...

I don't know anything about Bluesound, but the Sonos is so mature in terms of UX I can highly recommend it. I don't use music above 320k on it, but I only have a Play:3 (for now - I can feel a Connect coming on).

Oh, I like the look of Sonos. But I'd like to know if they can play the higher res files. I don't particularly care if they are downsampled in the process, but I read they need to be converted first which is a pain.

The other option of course is a media centre mini-PC which could do anything at about the same price!

This caught my attention...

https://www.asus.com/ASUS_VivoPC/

 

[johngw]

The other option of course is a media centre mini-PC which could do anything at about the same price!

This caught my attention...

https://www.asus.com/ASUS_VivoPC/

That looks very good. More stylish than the Intel NUC, cheaper also?

I've got a Mac Mini in the main system and opted for a Shuttle PC for the 2nd which I regret. It's bulkier than I had expected and was also very noisy until I replaced the case fan, which wasn't straightforward as it's not standard size.

 

[fr0g]

The other option of course is a media centre mini-PC which could do anything at about the same price!

This caught my attention...

https://www.asus.com/ASUS_VivoPC/

That looks very good. More stylish than the Intel NUC, cheaper also?

I've got a Mac Mini in the main system and opted for a Shuttle PC for the 2nd which I regret. It's bulkier than I had expected and was also very noisy until I replaced the case fan, which wasn't straightforward as it's not standard size.

Many people would be surprised to hear I was thinking about a Mac Mini too. But unless I go second hand (which I don't do), they are far too overpriced.

I fancy the low power of the Asus too, especially as there will be no screen anywhere near it other than for setting up.

 

[matt49]

The other option of course is a media centre mini-PC which could do anything at about the same price!

This caught my attention...

https://www.asus.com/ASUS_VivoPC/

That looks very good. More stylish than the Intel NUC, cheaper also?

I've got a Mac Mini in the main system and opted for a Shuttle PC for the 2nd which I regret. It's bulkier than I had expected and was also very noisy until I replaced the case fan, which wasn't straightforward as it's not standard size.

The thing that exercises me most about streaming music is the userfriendliness of the control software. Sonos is brillant. I find iTunes (this is on an iPad using Apple Remote). I mean, it works but it doesn't do a lot of the stuff I want it to do. The native app on my Synology NAS is just about OK, but compared to Sonos it really is a bit carp. With some Synology models you can replace the streaming software with other programs, but not mine, the DS213j.

I feel a new thread coming on ...

Matt

 

[pauln]

The other option of course is a media centre mini-PC which could do anything at about the same price!

This caught my attention...

https://www.asus.com/ASUS_VivoPC/

That looks very good. More stylish than the Intel NUC, cheaper also?

I've got a Mac Mini in the main system and opted for a Shuttle PC for the 2nd which I regret. It's bulkier than I had expected and was also very noisy until I replaced the case fan, which wasn't straightforward as it's not standard size.

Many people would be surprised to hear I was thinking about a Mac Mini too. But unless I go second hand (which I don't do), they are far too overpriced.

I fancy the low power of the Asus too, especially as there will be no screen anywhere near it other than for setting up.

How about a case like this http://www.bit-tech.net/hardware/cases/2013/09/12/akasa-newton-v-review/1 and build it yourself?

 

[johngw]

How about a case like this http://www.bit-tech.net/hardware/cases/2013/09/12/akasa-newton-v-review/1 and build it yourself?

Looks much more attractive than the standard Intel NUC plastic box, but at the price of £48 plus £310 for the motherboard, adding CPU, memory and disk, you're in Mac Mini territory from a price perspective... Likely above even.

 

[johngw]

Many people would be surprised to hear I was thinking about a Mac Mini too. But unless I go second hand (which I don't do), they are far too overpriced.

I fancy the low power of the Asus too, especially as there will be no screen anywhere near it other than for setting up.

If your media centre mini PC is going to be used headless, why not use any old (cheap) machine, hide it in a cupboard, and stream via AirPlay to an AEX next to your HiFi?

 

[steve_1979]

In order to move this dicussion onwards we need to find a way to look at this issue in a sensible and productive manner.

This thread has thrown up the following 'facts'.

Nyquist-Shannon states quite clearly that Red Book standard processing will recreate the original waveform exactly, I am comfortable with that.

It is also clear that certain other processes need to be carried out, I am thinking of anti-aliasing filters and noise shaping primarily but there are others.

These processes have a measureable effect and increase noise and distortion. A greater bit rate and sampling frequency will reduce these effects.

Therefore it is clear that 24/96 playback will be superior to 16/44.1 in terms of measurement. Again I have no problem with that.

Many people (myself included) have tried various tests and found it very dificult to distinquish 24/96 from 16/44.1 all other factors being equal.

Put simply, noise and distortion is present in greater quantity in 16/44.1 than it is in 24/96 but is inaudible to many listeners.

So I was looking at a way that this difference could be correlated and perhaps measured. I have heard the phrase 'effectively transparent' used in similar contexts.

So if we were to measure distortion and noise, is there a figure that could be agreed on, below which these distortions are inaudible and the process (or component) deemed effectively transparent?

Great post!

 

[johngw]

In order to move this dicussion onwards we need to find a way to look at this issue in a sensible and productive manner.

This thread has thrown up the following 'facts'.

Nyquist-Shannon states quite clearly that Red Book standard processing will recreate the original waveform exactly, I am comfortable with that.

It is also clear that certain other processes need to be carried out, I am thinking of anti-aliasing filters and noise shaping primarily but there are others.

These processes have a measureable effect and increase noise and distortion. A greater bit rate and sampling frequency will reduce these effects.

Therefore it is clear that 24/96 playback will be superior to 16/44.1 in terms of measurement. Again I have no problem with that.

Many people (myself included) have tried various tests and found it very dificult to distinquish 24/96 from 16/44.1 all other factors being equal.

Put simply, noise and distortion is present in greater quantity in 16/44.1 than it is in 24/96 but is inaudible to many listeners.

So I was looking at a way that this difference could be correlated and perhaps measured. I have heard the phrase 'effectively transparent' used in similar contexts.

So if we were to measure distortion and noise, is there a figure that could be agreed on, below which these distortions are inaudible and the process (or component) deemed effectively transparent?

Great post!

Actually, to nitpick, the above in bold is false. Depends on what you mean with "the original waveform". But I agree with the rest...

 

[steve_1979]

What we could also do is take the 24/96 track and the 16/44.1 conversion and extract the difference as a WAV file. Play that back...

Do you hear anything?

If you hear something here, then it is "possible" you will hear a difference between the tracks. If you don't, then it is quite clearly an inaudible difference.

Something I might try at the weekend.

Great idea! Now that should bring about some good quality evidence (either for or against) whether any differences can be heard between high rez and 16/44.1.

I'll be very interested to hear your results fr0g. I'll give this test a try myself sometime soon too but I'm going to be too busy this weekend.

 

[johngw]

What we could also do is take the 24/96 track and the 16/44.1 conversion and extract the difference as a WAV file. Play that back...

Do you hear anything?

If you hear something here, then it is "possible" you will hear a difference between the tracks. If you don't, then it is quite clearly an inaudible difference.

Something I might try at the weekend.

Great idea! Now that should bring about some good quality evidence (either for or against) whether any differences can be heard between high rez and 16/44.1.

I'll be very interested to hear your results. I'll give this test a try myself sometime soon too but I'm going to be too busy this weekend.

Agreed. Obviously the difference would have to be described in 24/96 - no way to effectively diff the 2 files otherwise - so you would downsample 24/96 to 16/44.1 (with dithering etc applied), then upsample to 24/96 again, then compare against the original.

 

[fr0g]

Many people would be surprised to hear I was thinking about a Mac Mini too. But unless I go second hand (which I don't do), they are far too overpriced.

I fancy the low power of the Asus too, especially as there will be no screen anywhere near it other than for setting up.

If your media centre mini PC is going to be used headless, why not use any old (cheap) machine, hide it in a cupboard, and stream via AirPlay to an AEX next to your HiFi?

1. I hate my work iPhone 5. I find iOS the most cumbersome, awkward OS I have ever used. Thank the lord of science I also have my private Galaxy Note II which beats the iPhone 5 into a bloody pulp of irrelevance in almost every way. I use a Windows 8 tablet as it takes a 64 GB SD card and runs all of my old software... Airplay is out of the question.

2. I intend having zero furniture other than my beautiful ADM9RS (walnut),stands, subwoofer and playback device... A tower PC would rather detroy the Feng Shui

I would buy a Mac Mini if it was under £400. It isn't. In fact here, it would be around £550. The Asus I linked looks as good, and would do everything I needed.

 

[steve_1979]

2. I intend having zero furniture other than my beautiful ADM9RS (walnut),stands, subwoofer and playback device... A tower PC would rather detroy the Feng Shui

Did you get those lovely looking Dynaudio stands in the end?

 

[steve_1979]

Duplicate post. :doh:

 

[fr0g]

2. I intend having zero furniture other than my beautiful ADM9RS (walnut),stands, subwoofer and playback device... A tower PC would rather detroy the Feng Shui

Did you get those lovely looking Dynaudio stands in the end?

No Steve, more's the pity... that will be a future 'upgrade'

seems illogical just now...

 

[the record spot]

Thankfully, and the more I read of this thread, the more I am grateful, I'll have a life this weekend.

A straight compare of Tannoy DC4, some Mission 761i and my new Pioneer S71B speakers.

Ah...