Skip to main content

Sorry if this has been discussed.

Just curious.What bit depth/SR and why?I just Read the interview in TapeOp (Marc Aubort, nice article!) and it got me thinking..He still records at 16 bit. Ethan Winer does too, if I recall correctly.

I see the studio guys discussing this a lot, but would like to see what you guys are working with. Ive been really curious about this subject for a long time.

I normally record at 24/96 because it seems to sound much better to me(but heck, what do I know, I think schoeps omnis sound like they are inside a paper tube :P) I also create DVD-Audio discs from all of my projects for later listening in my newly built playback room.

Comments

BobRogers Thu, 08/10/2006 - 06:58

I can well understand how someone good could record at 16 bits with no trouble. For a hack like me, 24 bits are like training wheels. I can leave myself a lot of headroom and still have a pretty reasonable signal to noice ratio. For sample rate, I usually just use 48. For me to hear the difference between 48 and 96 I have to listen to the recording rather than listen to the music if you know what I mean.

sheet Thu, 08/10/2006 - 08:17

For my music and application (everything I do goes to Avid eventually) I have to use 24/48.

The reason that people hear a difference in sample rates has long been debated. It is NOT because of the extended frequency response. That is insane. Most microphones, many mic cables and some low-end electronics do not have extended, linear range out to 44.1k, 48k, 96k, 192k, etc. We could not hear those frequencies lienearly anyway. It would require an enormous high f shelf to make those audible. Add that enormous shelf, and you are going to add distortion and bring out some nasty digital artifacts. It is easier for digital to represent bass than highs.

Since we cannot capture it, we are not recording it, we are hearing some other artifact of increased sample rates. The REAL issue is the bit depth. Increase the dit depth, get a better representation of the wave.

If I show you high res pictures at a rate of 44,100 times per second on a computer monitor calibrated for 256 colors, or 96,000 pictures per second of the same pic, on the same monitor, will that increase the resolution of the picture? Will that make it more realistic? No. If I change the monitor calibation to one million colors, then you have a better idea of what the high res pic actually is. Then you can determine when you are at the point of diminishing returns by taking up additional storage space on your drive, storing multiple copies of the same pic. That is what you must do with sample rates and audio.

While 16 bit is fine for some, I hear a difference between 16 and 24.

RemyRAD Thu, 08/10/2006 - 09:34

I'm one of those folks that if given the opportunity to record at 24-bit, I will but usually, I don't! Most of the recordings I make sound perfectly wonderful that people whether I'm rolling at 16-bit 44.1kHz or 24-bit 44.1kHz, so I usually stick with 16-bit. Sure there is a difference in the sound but I generally don't care about that kind of minute nuance. If you are a good engineer, you'll have no problem working within the constraints of a 96 DB dynamic range. I mean heck! In the good old analog days, we only had between 50 and 65 DB usable dynamic range and that never kept any good engineer from producing good hits. It would be much better to own a Rolls-Royce over a Chevrolet but a Chevrolet will still get you from point a to point B. So will 16-bit.

Practically practical
Ms. Remy Ann David

Cucco Thu, 08/10/2006 - 10:23

Well, it's been a while since a question like this popped up around here, so why not.

Usually - most projects go 24/44.1 (though I record into Sequoia at 32 bit, the last 8 bits are actually 00000000)

On higher-end projects, I'll go 88.2. On projects which may eventually go to high-res surround, I'll go 176.4. The reason for the 44.1, 88.2, or 176.4 is mainly b/c there is negligible difference between each of these and their 48kHz related bretheren and they save that little bit of hard-drive space. (Though I'm not usually worried about HD space, it's nice to conserve where one can.)

I don't use the 24 bit crutch for the extra s/n ratio but for the possibility of adding effects later. When I add a reverb, eq or minor limiting, it just sounds better when I have a higher resolution file to work with.

Though I'm finding myself using a LOT more outboard lately, so it ain't that much of a big deal.

J.

anonymous Fri, 08/11/2006 - 09:10

sheet wrote:
If I show you high res pictures at a rate of 44,100 times per second on a computer monitor calibrated for 256 colors, or 96,000 pictures per second of the same pic, on the same monitor, will that increase the resolution of the picture? Will that make it more realistic? No.

This analogy doesn't really apply because the values you're using aren't comparable. But, it is incorrect anyways.

Replace "times per second" with Hertz. Then go to you monitor settings and actually do this...granted you won't be able to select the same settings, but set it to your lowest frequency then set it to your highest. "See" the difference? Does the image on screen look more realistic?

I agree that bit depth is more important than SR, but once you've got the highest bit depth why not go for better SR frequency. The object of PCM recording is to make it sound as accurate as possible right?

I record at 88.2/24bit, but if I had the Horsepower/Storage I'd record everything at the highest possible resolution/format all the time. How about having 24 channels of DSD recording? Its only going to keep getting better.

Is the difference between 44.1 , 96 , 192 Night and Day?....well no, but there is a difference. I can see why some people don't really care about SR or dont have the most recent gear to use high SR(myself incuded), but to state that there isn't a difference between SR's is ridiculous.
I can't understand why we still have this debates over SR. The technology is there for a reason and it is well documented. This simplest way to tell the difference is just listen to the different SR's. Its not a conspiracy to make all your gear obsolete. (though if it was its working).

My ASSumption is that cds will be the last popular media for music distribution. Then what happens when the average Joe's internet connection is so fast that he can download/stream high resolution Audio/Video. BYE BYE format restrictions, file size restrictions, record companies.....ok, back to earth.

FifthCircle Fri, 08/11/2006 - 09:44

24 bit 44.1 for all my concert work. 24/96 for commercial projects. I use R8brain for SRC.

Jeremy- unless you're doing destructive calculations to your audio, you aren't gaining anything by recording at 32 bit. It is a legacy bit rate in Sequoia from the days before they figured out how to make a PC record a 24 bit fixed point waveform. Even if you are doing destructive work, you can save your work as new files and you can have the best of both worlds.

--Ben

Cucco Fri, 08/11/2006 - 16:39

FifthCircle wrote: 24 bit 44.1 for all my concert work. 24/96 for commercial projects. I use R8brain for SRC.

Jeremy- unless you're doing destructive calculations to your audio, you aren't gaining anything by recording at 32 bit. It is a legacy bit rate in Sequoia from the days before they figured out how to make a PC record a 24 bit fixed point waveform. Even if you are doing destructive work, you can save your work as new files and you can have the best of both worlds.

--Ben

The main reason I actually record in 32 bit is simple - I bounce all of my pre-masters out into 32 bit (which will make a difference when using effects or making level changes). I know that supposedly the Bit Depth Change from 24 to 32 should incur no issues or artifacts, however, I don't trust it unless I can see it.

SOooooo...I do 32 all the way around.

Personal preference.

anonymous Fri, 08/11/2006 - 17:46

Thanks for all the input folks, keep em comin'.

Also,I sometimes record in DSD and then convert from DSD to PCM....I dont know why, but PCM that has been converted from DSD *seems* to sound better than PCM that has stayed thus the entire time. Maybe I am losing my mind.

FifthCircle wrote: I use R8brain for SRC.

Here too(well, R8brain pro, though I assume that is what you use too, Ben?)

JoeH Fri, 08/11/2006 - 20:38

For quite a long time now, we've settled into 24/44 here, for all the inputting, DSP and master stereo mixes (usually put away for safe-keeping and possible DVD-A work).

Then of course, we bounce/dither down to 16/44 for CDs like everyone else.

I've been telling people for years that 16/48 or 48-ANYTHING was a waste of resources (Unless you're doing broadcast and video, of course), but the myths persist - no thanks to all the digital consoles and consumer DAT salesmen who tell people "48K is higher, so it's better!".

I still believe that any percieved advantages of going at 48K is lost in the gearboxing process back down to 44 for CD work, and as Sheet has already pointed out, the bit-depth is far more important than the sample rate. When resources permit it, it's nice to do 24/88 or 96, so going down to either rate - 44 or 48 - usually works out fine.

I will also admit that it's far less worrisome to go UP to 48K for most video soundtracks, if we started at 44k. I think most would agree that all but the most fanatic home videophiles don't have a problem with the soundtrack of most DVDs. (It's that extra distraction factor - the picture itself - that seems to keep most folks otherwise occupied.) When making an AC3 Dobly soundtrack, well, we all know what happens to those lost bits and rear speakers, eh? :roll:

Conversely, I think it's human nature (esp when listening with NO picture present) to be able to focus more on the sound alone, so audio folks are much more tweaky about it all, even with 16/44 CDs.

I'm VERY VERY happy with 24/44 in my day to day work here, and no one, I mean NO ONE, not one single paying client, has ever complained, or asked for anything higher.

FifthCircle Sat, 08/12/2006 - 11:28

TeddyBullard wrote:
Here too(well, R8brain pro, though I assume that is what you use too, Ben?)

Yes, of course...

cucco wrote:

The main reason I actually record in 32 bit is simple - I bounce all of my pre-masters out into 32 bit (which will make a difference when using effects or making level changes). I know that supposedly the Bit Depth Change from 24 to 32 should incur no issues or artifacts, however, I don't trust it unless I can see it.

SOooooo...I do 32 all the way around.

Interesting thought process... In my experiences, 24 vs. 32 bit recording makes no difference. Digital zeros are digital zeros. It doesn't change the way that things sum inside the workstation. As I said before, the only place where I've noticed that it can make a difference is if you are doing destructive (offline) processes (especially FFT or denoising). Even then, if you tell Sequoia to make new files, you'll get the benefit.

--Ben

--Ben

anonymous Sat, 08/12/2006 - 15:11

Teddy

I would be interested in some more details as to how you actually do this.
Thanks

Larry Elliott

TeddyBullard wrote:
Also,I sometimes record in DSD and then convert from DSD to PCM....I dont know why, but PCM that has been converted from DSD *seems* to sound better than PCM that has stayed thus the entire time. Maybe I am losing my mind.

anonymous Sat, 08/12/2006 - 18:23

either TASCAM DV-RA1000->Analog XLR Out->Mytek Stereo192 A/D>>

Or Via the new Discwelder Bronze software which converts from Raw DSD(dff) to PCM. Came free with the DVRA1000.

I dont know the reasoning behind why it sounds better??maybe psychological?

lell010 wrote: Teddy

I would be interested in some more details as to how you actually do this.
Thanks

Larry Elliott

[quote=TeddyBullard]
Also,I sometimes record in DSD and then convert from DSD to PCM....I dont know why, but PCM that has been converted from DSD *seems* to sound better than PCM that has stayed thus the entire time. Maybe I am losing my mind.

Simmosonic Sun, 08/13/2006 - 05:06

TeddyBullard wrote: Just curious.What bit depth/SR and why??

Up until 2003 I was doing everything at 24/44.1k. I can definitely hear a difference between 16-bit and 24-bit, but more importantly, 24-bit gave me an extra safety margin because much of my recording then was live concerts of chamber music which benefitted from both the extra resolution and the extra headroom - less chance of an 'over'!

Back then I was using a Prism AD124 24-bit AD converter, a beautiful piece of equipment that I sold to a good friend who still hires it out regularly to local studios. It only goes up to 48k sampling rate...

In 2003 I sold the Prism (and the rest of that rig) to move on to the Nagra V, which is much better suited to my needs. It sounds great at 24/44.1k, and for my 'freebie' live concert recordings for radio broadcast I tend to use 44.1k. But for everything else I use 96k for two reasons...

Firstly, for future-proofing. I'd hate to make a great recording for someone and then, some time later, have them tell me they plan to release an audiophile version (DVD-A, SACD, TTCAPOS) and requesting the high resolution masters! Hard disk space is cheap these days and, when you work direct-to-stereo as I do, there is plenty of processing power in even the simplest contemporary PC...

Secondly, one of the true benefits of a higher sampling rate is in the anti-aliasing filters required on the input of the AD circuit. At 44.1k, these need to be *extremely* steep and that is hard to do in hardware without introducing all kinds of problems that can extend down (as far as an octave below the Nyquist) into the audio passband - hence the high prices of Prism and similar AD converters for many years. But at higher sampling rates you shift those problems out of the audio passband. When you SRC down to 44.1k the anti-aliasing filtering is still required but now it can be done digitally; it is much easier to achieve the required response without the problems. So, in other words, working at 96k and converting down to 44.1k can yield very good results from affordable gear.

I am happy to use either sampling rate, but given a choice I'll use 96k for the reasons given above. As for word size, 16-bit just doesn't work for me. I can use it if I have to, but I'd rather not because 24-bit means one less thing to worry about - 'overs'!

Oh, that reminds me. 24-bit recording integrates much more easily with analog gear, in my opinion. You can set a nominal operating level where -20dB FS or -24dB FS corresponds to an output voltage of +4dBu and that gives you 20 or 24dB of headroom available on the recording medium, so you're not creating a headroom bottleneck. Recording at such a low nominal level on a 16-bit recorder is not a good idea, in my opinion; you need to be up around -14dB FS or so, which severely restricts your headroom to 14dB. Okay for rock and pop where 'headroom' is a dirty word, 'dynamic range' is considered a problem and everything is compressed to the point of being dead, but it's not for me.

[Flack jacket on...]

anonymous Sun, 08/13/2006 - 05:32

Which DAW and SRC/dither do you use for redbook conversion?

Simmosonic wrote: [quote=TeddyBullard]Just curious.What bit depth/SR and why??

Up until 2003 I was doing everything at 24/44.1k. I can definitely hear a difference between 16-bit and 24-bit, but more importantly, 24-bit gave me an extra safety margin because much of my recording then was live concerts of chamber music which benefitted from both the extra resolution and the extra headroom - less chance of an 'over'!

Back then I was using a Prism AD124 24-bit AD converter, a beautiful piece of equipment that I sold to a good friend who still hires it out regularly to local studios. It only goes up to 48k sampling rate...

In 2003 I sold the Prism (and the rest of that rig) to move on to the Nagra V, which is much better suited to my needs. It sounds great at 24/44.1k, and for my 'freebie' live concert recordings for radio broadcast I tend to use 44.1k. But for everything else I use 96k for two reasons...

Firstly, for future-proofing. I'd hate to make a great recording for someone and then, some time later, have them tell me they plan to release an audiophile version (DVD-A, SACD, TTCAPOS) and requesting the high resolution masters! Hard disk space is cheap these days and, when you work direct-to-stereo as I do, there is plenty of processing power in even the simplest contemporary PC...

Secondly, one of the true benefits of a higher sampling rate is in the anti-aliasing filters required on the input of the AD circuit. At 44.1k, these need to be *extremely* steep and that is hard to do in hardware without introducing all kinds of problems that can extend down (as far as an octave below the Nyquist) into the audio passband - hence the high prices of Prism and similar AD converters for many years. But at higher sampling rates you shift those problems out of the audio passband. When you SRC down to 44.1k the anti-aliasing filtering is still required but now it can be done digitally; it is much easier to achieve the required response without the problems. So, in other words, working at 96k and converting down to 44.1k can yield very good results from affordable gear.

I am happy to use either sampling rate, but given a choice I'll use 96k for the reasons given above. As for word size, 16-bit just doesn't work for me. I can use it if I have to, but I'd rather not because 24-bit means one less thing to worry about - 'overs'!

Oh, that reminds me. 24-bit recording integrates much more easily with analog gear, in my opinion. You can set a nominal operating level where -20dB FS or -24dB FS corresponds to an output voltage of +4dBu and that gives you 20 or 24dB of headroom available on the recording medium, so you're not creating a headroom bottleneck. Recording at such a low nominal level on a 16-bit recorder is not a good idea, in my opinion; you need to be up around -14dB FS or so, which severely restricts your headroom to 14dB. Okay for rock and pop where 'headroom' is a dirty word, 'dynamic range' is considered a problem and everything is compressed to the point of being dead, but it's not for me.

[Flack jacket on...]

anonymous Sun, 08/13/2006 - 09:38

24/96.

Because I always make a stereo master at 24/96 on DVD-Video and a master on CD.

Because 24 allows me to be more conservative with the recording levels: at 24, I will not be obsessed about going as near as possible to 0 dB. -12 to -6 is fine.

Because 96 allows a degree of time alignment in post-production (with MS pickup and in B-Format surround for example) that is not possible if the original is at a lower SR.

- Kewl

Simmosonic Mon, 08/14/2006 - 05:25

TeddyBullard wrote: Which DAW and SRC/dither do you use for redbook conversion?

Well, I'm nowhere near as fussy about this as I used to be, partly because I think there is much less to be fussy about these days, and partly because when it comes to DAWs in particular people still seem to treat them like religion, i.e: "mine is the only right one, everyone else is wrong".

My travelling DAW consists of Wavelab running on an IBM ThinkPad, with numerous plug-ins, most of which I never use. For SRC I am happy with the Resampler (or whatever it's called) plug provided with Wavelab. The plug-ins I mostly use are the MS matrix provided with Wavelab (ToolsOne?), Waves Linear Phase EQ (for invisible EQ), Waves standard EQ (for visible EQ, like when I want to bring something forward in the balance), and sometimes a bit of compression/limiting. A real life-saver sometimes is Waves' S1 Imager + Shuffler - not only can it solve some positioning problems with direct-to-stereo recordings, the shuffler itself can be amazing when used carefully if you want/need to create a 3D effect. Another 'must have' is the Magneto tape simulator. That has saved my butt a few times, mostly when I've chosen small diaphragm condensers for contemporary string works and ended up with very thin or bright recordings. A touch of Magneto tames them nicely...

Another essential function of any DAW is the Delete key. Sometimes the best thing you can do for a bad recording is to process it through as many instances of the Delete key as are necessary to completely remove it from existence. Then apologise to the client and offer to do the recording again for free, but properly.

I purchased Wavelab on the recommendation of Mr Spearritt, back in 2003 when I was rationalising my entire recording rig and decided to move on from SADiE and its proprietary world (more bloody DAW religion). I've since seen Sequoia in use and its interface allows certain things to be done much faster (such as choosing fade curves and so on). If I was to choose again, I'd be considering Sequoia very seriously. But I'd be considering Pyramix very seriously too...

However, I am happy with Wavelab so I see no need to move away from it. Life is too short and, as with any major change in software or operating system these days, there is so much you have to teach it about yourself and how you like to do things that it is often not worth the effort. I've got work to do...

anonymous Mon, 08/14/2006 - 10:39

I liked wavelab 5 too, and only moved away from it because I needed a multi-track editor, so I bought Sequoia. I also love the WAVES s1. very nice tool. I dont have it anymore so I use MSED from Voxengo. Thanks again for the very nice reply! I love soaking up wisdom.

Teddy

Simmosonic wrote: [quote=TeddyBullard]Which DAW and SRC/dither do you use for redbook conversion?

Well, I'm nowhere near as fussy about this as I used to be, partly because I think there is much less to be fussy about these days, and partly because when it comes to DAWs in particular people still seem to treat them like religion, i.e: "mine is the only right one, everyone else is wrong".

My travelling DAW consists of Wavelab running on an IBM ThinkPad, with numerous plug-ins, most of which I never use. For SRC I am happy with the Resampler (or whatever it's called) plug provided with Wavelab. The plug-ins I mostly use are the MS matrix provided with Wavelab (ToolsOne?), Waves Linear Phase EQ (for invisible EQ), Waves standard EQ (for visible EQ, like when I want to bring something forward in the balance), and sometimes a bit of compression/limiting. A real life-saver sometimes is Waves' S1 Imager + Shuffler - not only can it solve some positioning problems with direct-to-stereo recordings, the shuffler itself can be amazing when used carefully if you want/need to create a 3D effect. Another 'must have' is the Magneto tape simulator. That has saved my butt a few times, mostly when I've chosen small diaphragm condensers for contemporary string works and ended up with very thin or bright recordings. A touch of Magneto tames them nicely...

Another essential function of any DAW is the Delete key. Sometimes the best thing you can do for a bad recording is to process it through as many instances of the Delete key as are necessary to completely remove it from existence. Then apologise to the client and offer to do the recording again for free, but properly.

I purchased Wavelab on the recommendation of Mr Spearritt, back in 2003 when I was rationalising my entire recording rig and decided to move on from SADiE and its proprietary world (more bloody DAW religion). I've since seen Sequoia in use and its interface allows certain things to be done much faster (such as choosing fade curves and so on). If I was to choose again, I'd be considering Sequoia very seriously. But I'd be considering Pyramix very seriously too...

However, I am happy with Wavelab so I see no need to move away from it. Life is too short and, as with any major change in software or operating system these days, there is so much you have to teach it about yourself and how you like to do things that it is often not worth the effort. I've got work to do...

Cucco Mon, 08/14/2006 - 11:16

Simmosonic wrote: I purchased Wavelab on the recommendation of Mr Spearritt, back in 2003 when I was rationalising my entire recording rig and decided to move on from SADiE and its proprietary world (more bloody DAW religion). I've since seen Sequoia in use and its interface allows certain things to be done much faster (such as choosing fade curves and so on). If I was to choose again, I'd be considering Sequoia very seriously. But I'd be considering Pyramix very seriously too...

If SADiE aggrevated you with its proprietary formats, you won't find Pyramix to be that much better. If you like recording in DSD/DXD, it very well may be THE DAW to choose. However, if the majority of the work you do is PCM, you'd probably be happier with the Sequoia system. Both are quite nice, but I did an exhaustive comparison of the two before dropping my hard earned $$ on them and came to the conclusion that unless I were willing to drop >$20K, the system posed no significant advantage to me. ($20K being - $7 for the software and the Mykerinos cards, $3K for a killer PC and $10 for either the DAD, Prism or Genex conversion). I simply couldn't justify that. I do, however, like the fact that I could use Cedar's stuff on Pyramix, but I'm content with being "stuck" with Algorithmix...

J.

Simmosonic Tue, 08/15/2006 - 02:31

Cucco wrote: If SADiE aggrevated you with its proprietary formats, you won't find Pyramix to be that much better.

Thanks, Jeremy...

I bought SADiE at a time when native processing still sounded pretty bad, and SADiE was one of the few systems that had its own DSP cards and so on, and would also allow you to burn a CD directly from the waveform editing window. And although I remained loyal to the platform for some time, eventually it really started to burn my toast that it just didn't want to 'play' with the Windows interface. Everything seemed like a chore, even loading tracks in from CD. Meanwhile, the native processing got better and better, processing itself got faster, and soon it was hard to defend SADiE, IMHO...

As for Pyramix, the main attraction for me is that one of the guys behind the company is an ex-Nagra engineer! Pretty dodgy criteria on my behalf, huh?!? Although I'd be hoping much of that Nagra 'thinking' came with him to Pyramix, in terms of reliability, etc. I DO like the fact that one of their metering options is an emulation of Nagra's modulometer - it might not seem like much, but I am so used to Nagra's modulometer now that I really enjoy working with it.

But for now, Wavelab is doing all that I need and I'm happy with it...