Does High Resolution Audio sound better

P.S. It occurs to me the experience of going from P.A. to real, then back to P.A. is not unlike going from CD to high res, then back to CD. ;-}

Don Hills · March 15, 2015

... If theory predicts there should be no significant audible difference between 24/96 and 24/192, I say we need a better theory.
I've long said I believe there is more to be learned about digital. ...

Thanks, Barry.

*I* say if there is a significant audio difference, it should be quantifiable and measurable. Granted, traditional metrics may not be measuring what makes the audible difference. Regardless, we need to be able to identify the differences so that we can (a) identify and correct the causes and (b) be able to generate known quantities of those differences so that we can identify the detection thresholds as has been done, for example, for jitter. We can then work towards keeping the differences well below the audibility threshold.

I agree that everything in the chain makes a difference, and every item needs to perform well above what is considered "transparent". For example, Benchmark manufacture broadcast audio distribution amplifiers. They have bandwidths of hundreds of KHz. Although this might seem far greater than required, they know that there will be multiple amplifiers in the chain between the source and transmitter. By the time you add up all the rolloffs, you are getting close to the audible band.

This is all a long-winded way of saying, you (personally) go to a lot of trouble to ensure that every link in your chain between the artist and the final product performs well above(below?) any threshold of audibility. Your use of 24/192 is consistent with that philosophy, regardless of whether it makes an audible difference or not.

EuroDriver · March 15, 2015

Hi Ricardo,

The answer depends of course on whom you ask. Some will say there is no loss. To my ears, I'd say that by -12 things are starting and by -24 they're well on the way. At lower levels, instruments just sound (to me) like much cheaper, less maintained versions of themselves, and the space is ill-defined.

Best regards,

Barry

Soundkeeper Recordings

www.soundkeeperrecordings.wordpress.com

Barry Diament Audio

Am I am on the right track thinking that at - 12 db the voltage/signal level is being roughly represented by 16 - 4 bits = 12 bits and at -24 db the voltage/signal level is being represented by roughly 8 bits?

bdiament · March 15, 2015

Thanks, Barry.
*I* say if there is a significant audio difference, it should be quantifiable and measurable. Granted, traditional metrics may not be measuring what makes the audible difference. Regardless, we need to be able to identify the differences so that we can (a) identify and correct the causes and (b) be able to generate known quantities of those differences so that we can identify the detection thresholds as has been done, for example, for jitter. We can then work towards keeping the differences well below the audibility threshold.

I agree that everything in the chain makes a difference, and every item needs to perform well above what is considered "transparent". For example, Benchmark manufacture broadcast audio distribution amplifiers. They have bandwidths of hundreds of KHz. Although this might seem far greater than required, they know that there will be multiple amplifiers in the chain between the source and transmitter. By the time you add up all the rolloffs, you are getting close to the audible band.

This is all a long-winded way of saying, you (personally) go to a lot of trouble to ensure that every link in your chain between the artist and the final product performs well above(below?) any threshold of audibility. Your use of 24/192 is consistent with that philosophy, regardless of whether it makes an audible difference or not.

Hi Don,

Agreed.

Happily, on that last item, the artists, their band members, myself, my colleagues who've heard them, and our customers all seem to feel it makes some very important audible differences.

Hopefully, one day we'll have the metrics that show this as unequivocally in terms of numbers and graphs.

Best regards,

Barry

Jud · March 15, 2015

Nicely said, Don.

bdiament · March 15, 2015

Am I am on the right track thinking that at - 12 db the voltage/signal level is being roughly represented by 16 - 4 bits = 12 bits and at -24 db the voltage/signal level is being represented by roughly 8 bits?

Hi EuroDriver,

At -12 is is more like 14-bits and at -24 it is 12-bits.

(You get about 6.02 dB per bit.)

Best regards,

Barry

EuroDriver · March 15, 2015

Hi EuroDriver,

At -12 is is more like 14-bits and at -24 it is 12-bits.

(You get about 6.02 dB per bit.)

Best regards,

Barry

Soundkeeper Recordings

www.soundkeeperrecordings.wordpress.com

Barry Diament Audio

Many thanks for this correlation. In the quiet passages what sort of levels is a solo instrument, say a flute playing at ?

bdiament · March 15, 2015

Many thanks for this correlation. In the quiet passages what sort of levels is a solo instrument, say a flute playing at ?

Hi EuroDriver,

This will vary of course, depending on the recording, the piece of music and perhaps a good number of other variables.

Even without going to the quiet parts, if the recording is made with the idea of preserving the dynamics of the original performance, and the peaks reach the top bit (i.e., they are at -6 dBFS or louder -- lets say the maximum peak reaches -0.3 dB, the loudest I'll ever let a peak get on a master, again, depending on the music), the music between the peaks (i.e., the average level) might be 14 dB lower for pop music and 20 dB or more for classical music. That puts the pop average at just under -14 dBFS and the classical at just under -20 dBFS.

With a 16-bit system, that puts the pop average at 14-bits and the classical average at 13-bits.

With a 24-bit system, the pop average would be at 22-bits and the classical average at 21-bits (both better than CDs best, even if was pedaling downhill with the wind behind it ;-}).

Now, that's just the average levels of sounds. Instrumental harmonics and spatial cues in the room might be down another 10 dB, 20 dB or more from there, perhaps lower with really wide-ranging program material left intact.

This is why I say thinking of CD as having a dynamic range (as opposed to a signal-to-noise ratio) of 96 dB is more than misleading. If distortion doesn't count, I would say maybe it *is* 96 dB... if distortion doesn't count. But I think distortion *does* count, in which case the real dynamic range before things get raggedy is (in my opinion) an order of magnitude less.

Best regards,

Barry

EuroDriver · March 15, 2015

Now, that's just the average levels of sounds. Instrumental harmonics and spatial cues in the room might be down another 10 dB, 20 dB or more from there, perhaps lower with really wide-ranging program material left intact.

Barry

Have you done any recordings at 24/384, and if so what were your impressions of the audible improvements over 24/192 ?

bdiament · March 15, 2015

Barry

Have you done any recordings at 24/384, and if so what were your impressions of the audible improvements over 24/192 ?

Hi EuroDriver,

The designer of my converters has created them to run at 192 as the maximum. I've often commented that when using them at 192 at recording sessions, this is the first recording device in my experience where I have not yet been able to discern the output from the direct mic feed. If the recording already sounds (to me) like my mic feed, I can't think of how it can get better. (My only complaint is that I can no longer blame the gear for any flaws in my recordings. Of those, I must take full ownership now. ;-} )

In view of how few converters I've found that can--to my ears--actually realize the potential of the 4x rates (176.4k and 192k), I am more than a little skeptical of claims for even higher rates. Sometimes I wonder sometimes if the claims of higher numbers are simply marketing in the absence of being able to claim how well they do lower rates. I note that such claims are not coming from any of the designers of the converters I've found to be the most transparent performers. They're all working at 4x rates.

Best regards,

Barry

Online Audible Dynamic Range Sound Test

semente · March 15, 2015

When I extracted my classical music from CD to HDD I analysed at least one track of every record with Audioleak and Spek and the average dynamic range was around 35dB while on a few it went almost as far as 60dB.

35dB is about 10-bit in resolution, but 35dB below maximum level is already very low in intensity, as can be experienced here:

At -48dB (8-bit) the sound is so low in level that I wonder how anyone can determine if the signal is high or low res...

R

semente · March 16, 2015

But if the maximum level of a recording is down by -6dB then the very soft passages of hat same pice with 35dB DR will not only have 9-bit of resolution.

semente · March 16, 2015

Ravel's Bolero, here by Dutoit and the Orchestre Symphonique de Montréal on Decca, might be a good candidate for testing Redbook's low resolution at low levels...there could be more modern recordings with better sound quality and/or wider dynamic range but this is the only one I have of the piece.

semente · March 16, 2015

This one looks like a better candidate:

Ravel: Boléro - LSO0693 - Hyperion Records - MP3 and Lossless downloads

bdiament · March 16, 2015

But if the maximum level of a recording is down by -6dB then the very soft passages of hat same pice with 35dB DR will not only have 9-bit of resolution.

Hi Ricardo,

If the max level is at -6 and the dynamic range is 35 dB, you're talking -41, which would be 10-bits, not 9-bit.

With a 24-bit format, that would be 18-bits. Again, better than Redbook at its very best.

This is what keeps cellos sounding like cellos instead of like kazoos. ;-}

Best regards,

Barry

Teresa · March 16, 2015

…It is one of the things I've found to jump out at me when comparing high res. vs. Redbook when both are from the same mastering…

Me too, when going from from 16/44.1 to high resolution PCM the whole soundstage opens up, both wider and deeper in addition the sound of the recording venue is considerably more pronounced.

…If theory predicts there should be no significant audible difference between 24/96 and 24/192, I say we need a better theory.

I've long said I believe there is more to be learned about digital.

I agree completely.

semente · March 16, 2015

If the max level is at -6 and the dynamic range is 35 dB, you're talking -41, which would be 10-bits, not 9-bit.

With a 24-bit format, that would be 18-bits. Again, better than Redbook at its very best.

Hi Barry,

I might be wrong here but if I'm taking 41dB : 6 = 6,8333 which is closer to seven...16-bits - 7 = 9-bits.

On a 24-bit file that would be 17-bits.

But that was my point exactly; there's more resolution on High-Res.

So the next question is: since the lowest level is down by 35dB, will I be able to hear that loss in resolution at normal listening levels?

I might just blow £5 and try this myself with Gergiev's Bolero recording, unless you have any "Kazzo-sounding" sugestions.

I won't just take your word for it.

Best,

Ricardo

bdiament · March 16, 2015

Hi Barry,

I might be wrong here but if I'm taking 41dB : 6 = 6,8333 which is closer to seven...16-bits - 7 = 9-bits.

On a 24-bit file that would be 17-bits.

But that was my point exactly; there's more resolution on High-Res.

So the next question is: since the lowest level is down by 35dB, will I be able to hear that loss in resolution at normal listening levels?

I might just blow £5 and try this myself with Gergiev's Bolero recording, unless you have any "Kazzo-sounding" sugestions.

I won't just take your word for it.

Best,

Ricardo

Hi Ricardo,

It isn't a matter of what is closer; it is a matter of how many bits are being used.

As to whether you'll hear it, that depends on your hearing acuity, your system and its setup, and perhaps most importantly, whether or not you are sensitive to what this does.

Keep in mind that this is not listening to something in isolation. The lower level components are present for sounds at all levels (if the recording is worth its salt). They are in the harmonics of voices and instruments and in the space surrounding the players.

All you need to have in order to find out if any of it is meaningful to you is multiple resolutions of the same recording, from the same mastering session.

Best regards,

Barry

http://www.soundkeeperrecordings.om

semente · March 16, 2015

As to whether you'll hear it, that depends on your hearing acuity, your system and its setup, and perhaps most importantly, whether or not you are sensitive to what this does.

Hi Barry,

Somehow, I knew you would say that.

Hopefully I won't be able to hear it and will live hapilly ever after with my CDs...

Keep in mind that this is not listening to something in isolation. The lower level components are present for sounds at all levels (if the recording is worth its salt). They are in the harmonics of voices and instruments and in the space surrounding the players.

I see; in fact it might be more obvious with a single source sounding playing at a much higher level than the others.

All you need to have in order to find out if any of it is meaningful to you is multiple resolutions of the same recording, from the same mastering session.

That is mandatory, I know, but hard to acomplish.

I have only downladed your samples of "Wind of Change", which doesn't have large dynamic swings...how about the other samples?

Best,

Ricardo

bdiament · March 16, 2015

...how about the other samples?...

Of course I can't say what you or anyone else will or will not hear. (And I'd be wary of anyone who claimed they could.)

You might try the others on the same page. Or perhaps from another source - if you can be sure the different files are from the same mastering.

The sample does not have to have big dynamic swings. It can be anything, at any level.

Perhaps it might be easier to hear with certain types of instruments with complex harmonic structures (cellos, reeds, etc.).

Being familiar with the true sounds of the instruments can help but I don't believe this would preclude hearing differences.

Best regards,

Barry

Jay-dub · March 16, 2015

You could buy the CD on Amazon but the asking price is a bit of a rip-off for a used disc:

Amazon.com: Francois Couperin, Gerard Lesne, Ensemble Il Seminario Musicale, Gerard Lesne (counter-tenor): Office des Tenebres: Music

There's another one on eBay:

Gerard Lesne Couperin Tenebres Three Tenebrae Lessons for Holy Wednesday CD HR | eBay

Best,

Ricardo

The Harmonic Classics website says

Contact us by e-mail to purchase a CD, a Master Tape clone on CD-R (exceptional sound quality and original booklet) or a full libretto

with a link to their "info" email address. Presumably the Master Tape clone on CD-R would be the best way to buy the album.

semente · March 16, 2015

Of course I can't say what you or anyone else will or will not hear. (And I'd be wary of anyone who claimed they could.)

You might try the others on the same page. Or perhaps from another source - if you can be sure the different files are from the same mastering.

The sample does not have to have big dynamic swings. It can be anything, at any level.

Perhaps it might be easier to hear with certain types of instruments with complex harmonic structures (cellos, reeds, etc.).

Being familiar with the true sounds of the instruments can help but I don't believe this would preclude hearing differences.

I am quite familiar with the sounds of many acoustic instruments, but I am suspicious of most high-res vs. redbook samples available online...

Cheers,

Ricardo

bdiament · March 16, 2015

I am quite familiar with the sounds of many acoustic instruments, but I am suspicious of most high-res vs. redbook samples available online...

Cheers,

Ricardo

Hi Ricardo,

I can certainly understand. All too many comparisons and "shootouts" I've seen have the different formats at different levels and sometimes feature completely different masterings. This is exactly why I created the Format Comparison page.

Best regards,

Barry