Topic: 24bit mastering...worth it? maybe not?

[Nick's Picks]

I'm just getting into 24bit so I don't have a lot of experience with this.  I can say that when listening on a Denon2900 I can hear a difference between CDDA and the few 24 bit DVDA discs people have sent me.  The difference is not subtle.  I also had a Pioneer Elite top-of-the-line player on loan and that also revealed an improvement with 24-bit sources.

Nick, I understand what you are getting at.  If you have very high quality upsampling DAC and transport, then you are likely to get a very big improvement in CDDA playback.  That improvement may push CDDA close to 24-bit quality.  Likewise a low resolution system will mask the gains of 24 bit over CDDA. 

But for those of us in the mid range gear (multiformat players in the 1k-2k retail range) I would think that the benefits of 24/96 are going to be more apparent.

If you are arguing that the overall hassle and cost of recording/authoring at 24-bit and having to convert for portability is more than the value you percieve with 16-bit sources and very good upsampling, well I don't think anyone can dispute that.  V3->JB3->CD is about as easy as it gets and for the money you spend on a laptop or 24-bit recorder, you can get a really sweet upsampling DAC.

I think its just my ears man.  they must be beat. 
I dont think of my system as anything lavishly high-end (it just sounds that wa)...i mean, my entire system almost fits into that price range for a dec.  Not that money = quality sound, but in the world of high end stereo gear, it can.
Its really all about what you want out of your recordings as an individual, and I guess the value and ease part of it ...as you put it above so perfectly, is where I lean.  I rave on about things like this because I think it can be overlooked at times.  I know I do, so I need to remind myself that you can do soo much with so little in terms of our litle hobbie. 

[Karl]

one more thing, when I started taping 24 bit with my V3 stepping up from a USBPre (also @ 24 bit), I noticed an increase in low level resolution. What did this translate into on playback? I could REALLY hear how crappy the rooms and PA systems I was taping were, and also I could hear so many more details of the chatter and conversation going on around me!! So maybe for most rock shows we go to the benefits of 24 bit could be debateable.

But for me personally the real benefit of 24 bit is for stealthing relatively unamplified music.  As those of you out there who record real instruments in real space know, live instruments can have a huge dynamic range. So in these situations by running at 24 bit, I can set my levels very conservatively, which gives me peace of mind during the show since I know I am not clipping. I hate to be thinking about levels rather than the music during the show, and it drives me nuts to be worrying about clipping when I am not able to check levels!! Afterwards, using the Waves L2 plugin, I can boost the levels significantly and still have the recording come out well due to the increased low level resolution of 24 bit. Of course this benefit is specific to the types of shows I like to tape, but something to consider...


Ben

Yeah, I was going to say pretty much the same stuff as Ben.

[scb]

i'm not saying that there is not a benefit to be gained from mastering at 24bit.  I can hear it.  Its there.

What i'm saying is that I've focused more out of playing 16bit w/my stereo, and it sounds as good as raw 24bit files.  Only cheaper and MUCH less PITA.
ie: a good CD player that upsamples to 24bit
a good DAC that upsamples to 24bit
some sort of digital processer that re-clocks at 24bit (like the monarchy audio products).
you know, that little sucker could prove to be the best thing for us.  A $250 box that kills jitter and re-clocks all incoming digital to 24/96...then on to your DAC from there...could be the easiest soluction and a sonic equal to those wanting to go for more.

I have no doubt that 24bit is better.  I supose from the archival standpoint, mastering at 24bit is the thing to do.  I just look at it from the listening standpoint, and I feel that the technology here does just as good a job souping up 16bit PCM as recording it at 24bit does in the first place

Scott Brown..
remember your recent comment on the V3 regarding 24bit sound, is it dissapionting because ANSR sounds so damn good, that the line is thinned that much?.

first, hope i don't come off as lecturing, but upsampling to 24 bit is not correct.  upsampling would be to 96 or 192khz.  changing bit depth has nothing to do with sampling.  i know you know this nick, i'm just being a pain in the ass about terminology because i see people say it all the time. 

anyway, about the v3.  i love it at 16 bit.  i think it sounds great.  i don't really think the v3 sucks at 24 bit, I just think that other boxes sound better at 24 bit.  i ran an ad2k at 24 bit a few times in summer '02 and prefer that sound.   I now have a mytek stereo192 and think it sounds better than the v3 at 24 bit too.   so in saying I'm not happy with the v3 at 24 bit, I just mean that if I were going to record in 24 bit all the time, i'd rather run a different a/d.  the only problem is i'd want to keep the v3 preamp, so that gets expensive...

[Tim]

thanks Scott, I was just parroting what all of the materials from Bel Canto said. What would the correct term be then for changing bit depth?

[scb]

i have no idea.   i guess i'd just say upconverting or something, but is that even a word?

and another thought.  nick's saying that he's got great cd playback so he can't really hear much of a difference between the great cd playback and standard dvd-a playback.

well what if you have a sweet dvd-a player?

[Tim]

thanks Scott!

[Brian]

After reading through this thread I got to thinking. Are we really recording/listening to true 24bit waveforms through machines advertised as such?  What's the highest dynamic range spec you have ever seen on an A>D or D>A.  here are some examples of converters that i could find.

Apogee:
AD/DA 16x : A>D - 120dB = 20bits ; D>A - 118dB = 19.667 bits
Rosetta 200/800 : 114dB = 19 bits
MME : 105dB = 17.5 bits
(source: http://www.apogeedigital.com  all specs A-weighted)

Benchmark:
ADC104/DAC104 : A>D - 116dB = 19.333 bits ; D>A - not given (assumed to be around 116dB)
ADC1/DAC1 : A>D - 120dB = 20bits ; D>A - not given but judging from tests i'd say around 115/116 dB = 19 bits
(source: http://www.benchmarkmedia.com All specs are A-weighted)

Mytek:
8x96 : A>D - 120dB(A-weighted) / 117dB Total = 20 bits/19.5 bits ; D>A : same as adc.
Stereo192 : I couldn't find a .pdf manual. Scott, could you provide the info? is it the same?
(source: http://www.mytekdigital.com)

and just for fun....some "16bit" machines

JB3 : could not find anything.  anybody have the info?  I saw a S/N Ratio of 96dB but that really doesn't give us much.
M1/D100/D8/D7: "Better than 87dB" = 14.5 bits
DAP1 :  Better than 90dB = 15 bits
DA20mkii : "91dB or greater" = 15.1667 bits
PCM-R500/700 : not mentioned in manual but has a S/N ratio of only 90dB
(soure: from the manuals here.   could not find a jb3 manual or dynamic range spec )

so it's obvious the notion of "16bit" & "24bit" are marketing terms but i guess that has always gone without saying As far as I can tell it is impossible for humans to make machines that can perform true 24bits of data at this point in time. Say it does happen though in the future. What will they call it?  32 bit?

and that's just bit-depth!  sampling frequency of the recordings compared to the the frequency response of your playback will determine what you might here.

I think the recording technology is pretty much there, but not as advertised.  still......120dB of dynamic range is just awesome.  Is the playback technology there for it though?  who knows?  Some say they hear the difference while others don't.  As for myself i can definitely hear a difference with the higher-end converters and playback.

just thinking out loud i suppose.......

Brian

[jk labs]

A most fascinating topic!

The "theoretical limit" for the dynamic range (DR) of an ADC is # bits x 6 dB. I.e. a DR of 96 dB for a 16 bit converter. But you must subtract the quantization noise+dither and any unwanted noise. Often a smallish amount of DC offset survive the converter as well. So in reality you see the best 16 bits devices delivering 92 to 94 dB when measured. 

24 bits converters are different beasts. If you attempt to AD convert using a single 5V IC needing 4 Volts peak to peak of differential input signal in order to reach 0 dBFS you run into problems. To simplify the math assume you reach 0 dBFS with 2 Volts RMS on the input. Now divide 2 Volts RMS by 2 and repeat the process 22 times. You end up with the smallest RMS voltage the ADC chip needs to resolve in order to be a true 24 bit converter: (the number you found goes here).

"All in one" ADC chips contains quite a few internal devices. Several of these internal substructures have
noisefloors way above the 24th (LSB) bit. And then you have problems with 1/f noise (pink noise). So to get
specs for publishing you apply filters (and effectively say that some of the noise present simply isn't going to be counted).

The upshot is that current single-device ADCs claim to deliver a dynamic range of "typically" 117 dB
often under conditions not easily found in writing.

There are commercial 24 bit converters available today pushing 130 dB DR (according to specs). They are buildt differently, have more bulk and carry pricetags best left facedown. 

Jon

[Nick's Picks]

i have no idea.   i guess i'd just say upconverting or something, but is that even a word?

and another thought.  nick's saying that he's got great cd playback so he can't really hear much of a difference between the great cd playback and standard dvd-a playback.

well what if you have a sweet dvd-a player?

"digital processing" ?
" dither ", can go in either direction.  its still the same terminology.
I use resample because in the stereo gear world, that is the terminology.  I know we all know better.

I never mentioned DVD-A....but might as well as its just high-res PCM.
Its got nothing to do w/the player as I use it as a transport only.  In my tests, I've run high-res .wav files played off of a laptop and through the Edirol FA-101s digital output into the Musical Fidelity CD-PRE24s digi input.
Now a days...I run DADs > digi out > digi iin on my Sony 3000es (far superiour DAC), or FR-2 AES output > sony.
There is no "upsampleing" w/the Sony, and I can hear the difference.  24/96 PCM upsampled to DSD in the sony sounds fucking great.  Best digital i've heard.
With the MF box, which did re-clock and resample to 24/96, I could hear nothing between the 24bit masters and the dithered redbook version.  nadda.  nil.  Was it a weakness in the box, or a strength in the way it handled incoming 16bit PCM ? 

I"m about to re-introduce "upsampling" to my Sony w/the Monarchy audio DIP 48/96 upsampler and by doing so I expect the improvment I hear in raw 24bit material to vanish.

So, a great deck.  Like what?  The Dennon 2900 ?
I've not heard a truely sick high-end DVP / CD player.  But I know that all the good ones upsample redbook.

[dale]

of course the quality is only as good as the weakest link in the signal chain.
dat is not 24 bit,  cd is not 24 bit.

24 bit recording is providing a larger dynamic range.
then you must dither the audio to 16 bit for playback in cd format.
the quality of the algorythym  in this rewriting process is important. it can slur and color the recording.
if you record at 16 leave it alone.

the extra headroom provide by 24 bit reording are 
with the extra dynamic range you can record at a level which allows trancients (ie drums)  more headroom.
this extra range also translate to the lower level signals to have better resolution.

dale  long run audio

[Brian]

A most fascinating topic!

The "theoretical limit" for the dynamic range (DR) of an ADC is # bits x 6 dB. I.e. a DR of 96 dB for a 16 bit converter. But you must subtract the quantization noise+dither and any unwanted noise. Often a smallish amount of DC offset survive the converter as well. So in reality you see the best 16 bits devices delivering 92 to 94 dB when measured. 

24 bits converters are different beasts. If you attempt to AD convert using a single 5V IC needing 4 Volts peak to peak of differential input signal in order to reach 0 dBFS you run into problems. To simplify the math assume you reach 0 dBFS with 2 Volts RMS on the input. Now divide 2 Volts RMS by 2 and repeat the process 22 times. You end up with the smallest RMS voltage the ADC chip needs to resolve in order to be a true 24 bit converter: (the number you found goes here).

"All in one" ADC chips contains quite a few internal devices. Several of these internal substructures have
noisefloors way above the 24th (LSB) bit. And then you have problems with 1/f noise (pink noise). So to get
specs for publishing you apply filters (and effectively say that some of the noise present simply isn't going to be counted).

The upshot is that current single-device ADCs claim to deliver a dynamic range of "typically" 117 dB
often under conditions not easily found in writing.

There are commercial 24 bit converters available today pushing 130 dB DR (according to specs). They are buildt differently, have more bulk and carry pricetags best left facedown. 

Jon

fascinating indeed!  +t

130dB of dynamic range? wow

I couldn't find that spec for the Sony ES Series.....

[lds490]

Of course 24 Bit is better than 16 bit.

This thread reminds me of that Spinal Tap scene--"This one goes to 11."

Actually, I wish I understood some of the technical stuff better.

[Xpanding Man]

I still listen to casssettes on my single motor, dual deck that's never been cleaned or demagnetized.....what is 24/96 ? 




I'ze just funnin' .... fascinating topic for me , since i've often wondered where we'll draw the line with ever-increasing bit and sampling rates....FWIW, I know of no sweeter sound than the UV22 on my AD1000; until something explodes or melts i'm happy with 16/48

[Nick's Picks]

"of course 24bit is better than 16bit "

lets put things in context.
in a less than perfect acoustical situation with a very, very loud noise floor (like....say a concert in any venue outside of a church) and with a source that has *very limited* dynamic range on its own....like say a PA system of less than complete state of the art technology....
and then tell me why its so important to record at 24bits.

shit, for what we are doing , shockmouts are overkill.

Youve got little to gain recording jam bands at 24bit, imo.  most especialy if your CD player or DAC will upsample redbook to 24/96.  That is the killer.

I didn't mean this to start into a 16bt vs 24bt "which is better" thread.  We've all heard that 24bit sounds better, and it does.  It just gets to a point of overkill in terms of playback on our systems and the marginal acoustical environments we are recording in.
Let me put it more accuratly, its just overkill for me.  I've built a modest stereo system that does wonderful things with 16bit audio.  I've been recording at 24bit for years (off and on since 2001).  I've heard plenty of 16bit and 24bit versions of the same files on several builds of my stereo.  To me, its just easier to massage the playback end into sounding like real high-end digital than it was to actualy master and then worry about playback performance of 24bit audio.
I'm confident that for $700 I can put together a digital front end that will play 16bit audio to such a degree that it is virtualy impossible to distinguish from 24bit masters through the same gear.
To me..that is makes far more sence than worrying about 24bit recorders and high-end DVD-Audio players.

I know that some of you are lucky enough to record real instruments in good settings, and of cousre this topic doesnt' really apply.

[Brian Skalinder]

Of course 24 Bit is better than 16 bit.

Or at least it offers better resolution, detail, and dynamic range.  Not necessarily "better" depending on a whole slew of factors.

Edit:  Errr...what Nick said.  Must engage brain and read posts before responding...