[nickspicks]

I performed an intesting test yesterday.

preface:   I do some occasional location recording for a hobby and have built a reference system that can acurately dish out anything I throw at it (imo).


Ok, so the other day i  performed A-B-A testing with a 24bit 96khz master from a concert I taped a few weeks ago, and the downsampled, dithered 16/44 redbook standard CDs I burnt from this 24bit master .wav file.
My stereos front end consists of a Musical Fidelity CD-PRE24 which automaticly upsamples redbook CDs to 24/96 before analog conversion...and I figured this would be interesting to see if it is actualy worth the trouble to record at high resolution when my stereo does such a fine job making 16bit audio sound so nice (and 24bit-like).

I hooked up my laptop > firewire > Edirol FA-101 (see www.edirol.com)  toslink output to my MF cd-pre24 and played the 16bit version (resampled in wavelab and dithered via the apogee plug-in) through the transport  so that I could switch between sources via the remote from my listening position.

long story short...
the upsampled redbook source sounded very nice.....but the raw 24/96 files sounded better. If that was 100%, than the redbook version through the upsampling DAC was 85%. Noticible, but not "in your face" different. The 24bit file was clearer sounding, more precise, better detail in the articulation of each instrument. Bass notes were deeper, tighter and I could hear the attack "pluck" of the musicians bass strings vs. the note that was just "there" w/the 16bit file. It was strange, ...somewhat like I could hear the origins of the individual notes in the raw 24bit files vs. the appearance of the notes as they seem to just materialize out of the redbook source. The 24bit files sounded, in a few words..cleaner and more natural.

so is it worth it?
hmmmmm....it makes me think that with the right DAC, 16bit will hold its own against 24bit any day of the week. My MF box is pretty nice sounding, but its not the end all, be all in redbook playack...plus, it also reclocks all digital input through its own circuitry (which I dont believe would produce any audible artifacts, but i'd rather it didn't).
I'll have to take both sources over to my friends house to see if we can hear a bigger difference in his system (Benchmark DAC-1).
Unfortunately, my ears were quickly spoiled w/the 24bit testing at my house....and now I need to keep recording at this resolution in order to really connect w/the music in playback. that small difference between the sources became more apparent as the listening went on into the evening. It would have been nice to come down to the realization that my upsampling DAC could cut the mustard.

Testing equipment consisted of...
source:
dickey Betts Band  6-24-04
AKG c414 (DIN, hypers) > edril FA-101 (24/96) > firewire out > toshiba 3480ct laptop
- raw 24bit 96khz master .wav fles
- 16bit / 44.1khz redbook CD (dithered with an apogee plugin in wavlab) and resampled (also w/wavlab) from the 24bit master.

front ends:
24bit- toshiba 3480ct running Wavelab 4.0 for playback
firewire link from laptop to Edirol FA-101. Toslink output from here...
16bit - MF CD-PRE24

Musical Fidelity CD-PRE24 (preamp, transport and 2496 upsampling DAC)
McCormack DNA-1 power drive (185wpc)
Von Schweikert VR4 gen I loudspeakers
Gregg Stralley custom speaker cable and single IC.

www.nickspicks.com- product reviews- DIY's and FAQs- music lists

[Dan Driscoll]

long story short...
the upsampled redbook source sounded very nice.....but the raw 24/96 files sounded better. If that was 100%, than the redbook version through the upsampling DAC was 85%. Noticible, but not "in your face" different. The 24bit file was clearer sounding, more precise, better detail in the articulation of each instrument. Bass notes were deeper, tighter and I could hear the attack "pluck" of the musicians bass strings vs. the note that was just "there" w/the 16bit file. It was strange, ...somewhat like I could hear the origins of the individual notes in the raw 24bit files vs. the appearance of the notes as they seem to just materialize out of the redbook source. The 24bit files sounded, in a few words..cleaner and more natural.

I'm not surprised at all, this is pretty much what I would have expected. Keep in mind that there is much more actual musical information in the raw 24/96 file. When you convert to 16/44.1 you actually throw a lot of that information away. Upsampling is simply an attempt to recreate what was thrown away. A much better solution is to simply not throw it out to begin with. This is why I'm looking at DAD (Digital Audio Disks) or SACD/DVD-A discs with hi-rez stereo tracks, instead of CDs, whenever possible.

[tubesguy]

Perhaps I'm missing something in the original post, but you said the 24/96 was better, the redbook was maybe 85%, but you conclude that 16/44.1 could be just as good, given a better DAC? Huh?

This is not just an academic matter to me.  I'm thinking of digitizing my vinyl collection, now that 24/96 recording and burning to DVD is practical for home use, so I'm really interested in your test.  It's just that the conclusion seems to be at odds with the observed results.

Thanks for the interesting post - Pat

[nickspicks]

Perhaps I'm missing something in the original post, but you said the 24/96 was better, the redbook was maybe 85%, but you conclude that 16/44.1 could be just as good, given a better DAC? Huh?

This is not just an academic matter to me.  I'm thinking of digitizing my vinyl collection, now that 24/96 recording and burning to DVD is practical for home use, so I'm really interested in your test.  It's just that the conclusion seems to be at odds with the observed results.

Thanks for the interesting post - Pat

I mean that a better DAC/Transport than mine might come close to raw 24/96...thus negating the need to record at 24/96 to begin with.

[nickspicks]

I'm not surprised at all, this is pretty much what I would have expected. Keep in mind that there is much more actual musical information in the raw 24/96 file. When you convert to 16/44.1 you actually throw a lot of that information away. Upsampling is simply an attempt to recreate what was thrown away. A much better solution is to simply not throw it out to begin with. This is why I'm looking at DAD (Digital Audio Disks) or SACD/DVD-A discs with hi-rez stereo tracks, instead of CDs, whenever possible.

well....the thing is, it is a much bigger hassle for me to record at 24bit resolution....and I am curious to hear how well a job upsampling DACs perform against raw 24bit data.
I know that lots of information is thrown away, but that is a little harsh for a description.  Its not that simple.  I didn't simply truncate the extra 8bits here.

Its surprising though....just how close the two files end up sounding when one 2gb in size and the other only 700mb.  that is a lot of data missing, but you wouldnt' know it from what you hear.

[Dan Driscoll]

I mean that a better DAC/Transport than mine might come close to raw 24/96...thus negating the need to record at 24/96 to begin with.

It would be an interesting experiment to see how close you could get with a real top of the line DAC, because your MF CD-PRE24 is already a pretty nice unit. It's not like you're starting with a mass market $100 CD player.

Another question is what if you upgraded the 24/96 DAC/transport to the same degree? Would that restore the 85% ratio? Possibly make the difference greater?

[Dan Driscoll]

well....the thing is, it is a much bigger hassle for me to record at 24bit resolution....and I am curious to hear how well a job upsampling DACs perform against raw 24bit data.
I know that lots of information is thrown away, but that is a little harsh for a description.  Its not that simple.  I didn't simply truncate the extra 8bits here.

I know it sounds a little harsh and I certainly understand that it simply isn't a matter of dropping the 8 LSB's and there is also the matter of the sample rates. But you are making what is basically a low resolution copy from a somewhat higher resolution copy. The 24/96 master is already missing info, the conversion to 16/44.1 loses even more.

Its surprising though....just how close the two files end up sounding when one 2gb in size and the other only 700mb.  that is a lot of data missing, but you wouldnt' know it from what you hear.

The human mind is an amazing thing, it will fill in the missing material on its own. This is especially true for music that we have heard before, our memory will provide the misssing pieces..

I think an interesting experiment would be to make 2 masters, 1 at 24/96, the other at 16/44. Then compare the upsampled 16/44 master to the 24/96 master. I suspect the 16/44 would sound even closer to the 24/96 than in your original experiment, where the 16/44 is a copy of the 24/96.

[nickspicks]

I think an interesting experiment would be to make 2 masters, 1 at 24/96, the other at 16/44. Then compare the upsampled 16/44 master to the 24/96 master. I suspect the 16/44 would sound even closer to the 24/96 than in your original experiment, where the 16/44 is a copy of the 24/96.

I bet you're right.