Confused by all the bits

[BigH]

That's great for that album, but say I order up, I dunno, a Kate Bush (big fan here) FLAC puporting to be downsampled from the studio master. If it's not signed in blood by Kate hereself how are we to know it's not just a rip off (and rip-off) the CD?

If its recorded at 24bits why should they upsample the cd?

There are various test samples on sites like Linn, however some of the mp3 samples are not straight downsizing, I know someone did a test.

Which Kate Bush are you talking about?

 

[igloo audio]

In terms of high resolution (24/192) vs. CD (16/44), you have the usual conundrum of whether the format sounds different or does your DAC sound different given the bit depth and sample rate?

The first is a very simple test to do, you can download Adobe Audition (free for 30-days) convert a 24/192 track to 16/44, and then up-sample the 16/44 track back to 24/192 containing 16/44 information. You can then compare the up-sampled 16/44 file to the original 24/192 file by playing both tracks simultaneously, but with one track inverted, and quickly confirm whether the tracks null or not. I.e. they should cancel each other out? You will then find the files null to -96db or thereabouts, proving there is no benefit with 24/192 material, unless you have a HiFi system and ears better than Nyquist.

What if the DAC sounds different with 24/192 material?

This time you convert a 16/44 file to 24/192 and then A/B the two files. Alternatively, you can record the analogue output using an AD/DA converter like the Lynx Hilo @ 24/192 resolution, which is capable of -100db in the analogue domain (asynchronous USB / balanced analogue loops) and compare as above using Adobe Audition with one file inverted. Basically, all we are doing is recording the output differences from the DAC, deducting one from the other to see what is remaining, and then deciding a little more scientifically whether we are really hearing a difference or is it expectation bias or other.

Same for USB, S/PDIF cables etc etc.

At least then you can deduce whether it's worth the extra expense downloading 24/192 material, or indeed, spending lots of money on cables. Unless you are buying a newly engineered or re-mastered album, I'd wouldn't bother trying to replace your CD purchases with 24/192 purchases. But as a throwaway comment, if studios are recording and mastering in 24/96, DSD or other, I don't see any reason why these formats can't be passed-through to the consumer as a purchase option.

Peter

PS. Hope that didn't get confusing and I made no mistakes. It's late you know!

 

[CnoEvil]

In terms of high resolution (24/192) vs. CD (16/44), you have the usual conundrum of whether the format sounds different or does your DAC sound different given the bit depth and sample rate?

The first is a very simple test to do, you can download Adobe Audition (free for 30-days) convert a 24/192 track to 16/44, and then up-sample the 16/44 track back to 24/192 containing 16/44 information. You can then compare the up-sampled 16/44 file to the original 24/192 file by playing both tracks simultaneously, but with one track inverted, and quickly confirm whether the tracks null or not. I.e. they should cancel each other out? You will then find the files null to -96db or thereabouts, proving there is no benefit with 24/192 material, unless you have a HiFi system and ears better than Nyquist.

What if the DAC sounds different with 24/192 material?

This time you convert a 16/44 file to 24/192 and then A/B the two files. Alternatively, you can record the analogue output using an AD/DA converter like the Lynx Hilo @ 24/192 resolution, which is capable of -100db in the analogue domain (asynchronous USB / balanced analogue loops) and compare as above using Adobe Audition with one file inverted. Basically, all we are doing is recording the output differences from the DAC, deducting one from the other to see what is remaining, and then deciding a little more scientifically whether we are really hearing a difference or is it expectation bias or other.

Same for USB, S/PDIF cables etc etc.

At least then you can deduce whether it's worth the extra expense downloading 24/192 material, or indeed, spending lots of money on cables. Unless you are buying a newly engineered or re-mastered album, I'd wouldn't bother trying to replace your CD purchases with 24/192 purchases. But as a throwaway comment, if studios are recording and mastering in 24/96, DSD or other, I don't see any reason why these formats can't be passed-through to the consumer as a purchase option.

Peter

PS. Hope that didn't get confusing and I made no mistakes. It's late you know!

I assume you have done this....and if so, what are your conclusions?

 

[igloo audio]

Sorry for not replying sooner, it's been very busy.

Yes, quite a lot of time for pleasure but also for business.

It all started with setting-up systems for me and customers trying to ensure the systems were bit transparent, making sure the PC had the right sound and software settings, and not leaving everything to drivers and guesswork. It is much easier with a Mac, because iTunes and Core Audio will pass bit perfect audio with the right Audio Midi settings. And equally some DAC's (Weiss) can test bit transparency, which makes it a bit easier on a PC - until you get a bit more confident with JRiver, Foobar, ASIO and WASAPI drivers etc. Some DAC's are awkward, like Exposure's 2010S2 DAC, where you have to ensure the output is set to 24/96, but it sounds as impressive as many others costing several times its price, especially given the innately small differences you find in well designed DAC's.

The conclusion as outlined in the first paragraph is fact, and easy to replicate as described by using Abode Audition, but there are instances where 24/192 can and does sound very marginally different, which is probably more to do with the conversion algorithm used in Audition, Korg AudioGate etc.

The second paragraph is more variable, since most DACs use their internal clock and resample the input data. Therefore, it's very difficult to get a deep null or any null for that matter when comparing the recorded file (edited and cut precisely) to the original file, unless your DAC can perfectly synchronize to the imbedded S/PDIF, AES, or Toslink clock. Given this, you are simply better to listen in most instances, and evaluate whether you can hear a differences between transports, cables etc. Personally, I find it very difficult to call depending on the system. But converters do and can sound quite different in transparent systems even if they appear to measure perfectly on paper, especially if connected direct to active speakers when you're best to consider it as a preamp. I personally find relatively budget bookshelf speakers like the ProAc Tablette Anniversary are pretty good at telling the differences when using transparent amplification. However, in the scheme of things, it's all pretty minor, at least compared to your speakers, room and setup.

If your transport and DAC use an external clock, or an internal asynchronous USB clock arrangement as with the Lynx Hilo AD/DA converter, you can digitally loop through USB, S/PDIF, AES and Toslink and always capture the original bit data perfectly. This proves bit transparency in iTunes/Core Audio, and in these instances cables make no difference. Equally, you can obtain very deep nulls in the analogue domain approaching or bettering CD resolution (-100db obtainable), which in itself dismisses affects due to noise or other claims made for digital cables in the context of the Lynx Hilo.

Mind you, there are always exceptions to the rule, and I guess you could call this bad design :doh:

Btw, always buy a DAC because it provides the right level of functionality, and ideally with a high quality preamp (analogue inputs) and headphone output, rather than be overly concerned about the sound. As above, there are exceptions to the rule. One that comes to mind is the DAD AX24 (our reference), which is one of the most musical performers given its weighty and fluid bass/mid, and is used by Abbey Roads Studio, Bob Katz and many well known engineers that record in DSD.

Peter

 

[CnoEvil]

Peter, thank you for the comprehensive reply.

I have a Linn MDS playing through a MF AMS35i and Kef 205/2s, and I have certain Linn downloads at 3 different resolutions (320/16bit/24bit) eg. "A case of you" by Ian Shaw. This is a simple piece that lets you hear deep into the recording. As the resolution improves, there is also an improvement, especially in the subtle ambient detail.

I had always assumed this was purely down to the improved resolution, but Steve has showed it could well be down to using better mastering. For me, the jury is still out, as I think Hi-res 24 bit may well have advantages.

 

[igloo audio]

Hi CnoEvil

Some pulse for thought re better sound or not.

The image below is showing Kate Bush's album 'Fish People' and track one "Snowflake" as a 24/96 recording (note 48kHz) with one of 24/96 files inverted and mixed down to cancel-out the other 24/96 file to -144db in yellow.

The orange line shows a 24/96 file reduced to 16/44 and then up-sampled to 24/96 vs. the original 24/96 file, which show the 16-bit noise floor of -96db until you reach 22 kHz. These files to all intent & purposes will sound the same, unless your system and ears can beat Nyquist.

In red, I've followed the very same process as above, but this time using Korg AudioGate as the conversion tool. Notice the output below 2k reaching well within the audible range of -50db, which is still very subtle - I wouldn't want to A/B, but it would be hard to dismiss an audiophile claiming to hear a difference with this 24/96 version.

Obviously, this beggars the question, is it the format/resolution (not according to Nyquist), or just the tool and its algorithm used to record, up-sample and convert?

Peter

 

[NHL]

And still, reviewers rave if 384 khz sampling is supported, with an extra star rewarded. Sometimes it seems like most of the review is done with SACD material and almost nill redbook CD material.