48K vs. 44.1 - any audible difference?

[dansouth]

I suspect that the human ear cannot reliably distinguish between 44.1 and 48K given the same program, word size, and when no sample rate conversion has been applied. Has anyone ever tried a blindfold test?

 

If my suspicion is true, is there any value to recording at 48 or 96K vs. 44.1 or 88.2? 44.1 requires no format conversion for CD production. I've heard that DVD audio formats of 96 or 192K have been proposed, but so far they're nothing more than vaporware. A lot will change before these protocols are finalized, and they could eventually be killed by well-meaning bean counters. (Anyone remember consumer DAT?)

 

What are your thoughts on the issue?

[Philip OKeefe]

Dansouth said:

 

>>I suspect that the human ear cannot reliably distinguish between 44.1 and 48K given the same program, word size, and when no sample rate conversion has been applied. Has anyone ever tried a blindfold test?<<

 

Yes, I have listened to them both in a double blind test, and while I can perceive a slight increase in top end clairity, I feel that there is usually more harm done in sample rate conversion than the benefit of the slightly higher sample rates offer. Opinions DO differ wildly on this however.

 

As far as your AW goes, unless there's a good reason to run it at 48 KHz (such as using a BRC and ADAT's locked up to it) I'd suggest you run 24 bit @ 44.1 KHz. I think you will notice a MUCH bigger improvement in the sound quality at those settings than the difference between 44.1 and 48 KHz sample rates, given the same bit resolution.

 

??If my suspicion is true, is there any value to recording at 48 or 96K vs. 44.1 or 88.2? 44.1 requires no format conversion for CD production. I've heard that DVD audio formats of 96 or 192K have been proposed, but so far they're nothing more than vaporware.<<

 

Well, DVD audio will eventually support 96 KHz, and when that becomes the de facto palyback medium, then they'll be a benefit of running your recordings at that speed, but as long as CD is the playback medium of choice, I really don't see any benefit of running at 48 KHz. Maybe 88.2 would be better, since it's exactly twice the rate of 44.1 - sample rate conversion should be much easier for the converter device to do, and therefore, SHOULD be better sounding.

 

>> A lot will change before these protocols are finalized, and they could eventually be killed by well-meaning bean counters. (Anyone remember consumer DAT?)<<

 

Yes, I still have a "Save DAT" sticker around from the late 80's. The convept that really interested me back then was "Shareware" music. If everyone had consumer DATs in their homes, then a band could put out the music as "shareware" and ENCOURAGE copying, but offer "bonus Tracks" liner notes and packaging etc. to those that "registered". I'm surprised that someone hasn't tried that idea with Napster and MP3's in a more organised way, but I HATE the sound of MP3's, and wouldn't be interested in purchasing MP3 stuff - maybe I would be willing to "audition" stuff on MP3's and purchase the CD. I can't wait until higher bandwidth allows higher quality music to be transferred over the Internet at reasonable speed. When THAT becomes universally available, I think the record industry as we know it will REALLY change.

 

Just my opinions - YMMV.

 

 

Phil O'Keefe

Sound Sanctuary Recording

Riverside CA http://members.aol.com/ssanctuary/index.html

email: pokeefe777@msn.com

 

What are your thoughts on the issue?

[Jon Atack]

Practically speaking, if you're recording to digital and you know for certain that it will be mixed on an analog console, then record at 48 kHz.

 

Otherwise, record at 44.1 kHz to avoid SRC down the road.

 

Either way, use 24 bits, not 16.

 

Jon

[dansouth]

Originally posted by pokeefe777@msn.com:

As far as your AW goes ... I'd suggest you run 24 bit @ 44.1 KHz.

 

Phil, great minds think alike. I've been sticking with 44.1 for the past few years because I want to avoid sample rate conversion. I've run the AW4416 at 44.1K, 24 bit for every project thus far.

 

The 48K question comes into play because I have a Triton Rack, and the basic samples were recorded at 48K. When I use analog outs, this is no big deal, but if I want to use the optional ADAT out on the Triton Rack, it might be better to track at 48K. In this case, I'd hire a fully equipped mastering studio to do the rate conversion to red book specs.

 

A second issue is with the Alesis MasterLink, which can store mixes at 24/96. This doesn't make a lot of sense if my mixes are tracked at 44.1. What a pain! Why didn't the industry just stick to multiples?

 

Dan

[Alon]

The average joe probably cannot distingish between 44.1 vs. 48 k, though I suspect that plenty of pros can, though the difference is minor.

 

I never mess with 48, cause I don't like to down convert the audio later. Sounds better to me to start at 44 and keep it that way. Though I suppose it depends on the equipment you happen to be working with. If your gear only records at 48, then I suppose you're stuck with using that.

 

As far as Mp3s are concerned, I too am obviously not thrilled about the sound quality. Unfortunately, I don't think that the average joe really gives a shit about that.

[Lee Flier]

Where the sampling rate really matters IMO is when applying DSP. EQ, dynamics and other DSP is more responsive and sounds better when it's applied to more audio. Now the benefit there between 44.1 and 48 may still be negligible, but at 96K you definitely notice the difference.

 

I also have tracked everything on my AW4416 at 24 bit/44.1K so far, but I intend to re-cut a short piece at 48 just to see if the DSP is any more responsive. If it's that much better I can always record critical stuff at 48 and master it on my computer.

 

--Lee

 

This message has been edited by Lee Flier on 01-23-2001 at 08:27 PM

[alphajerk]

i strictly use 48khz. i notice a difference and it makes for a much more pleasurable mixing experience and the only time you have to hear 44.1 is on the burn. SRC has gotten pretty damn good lately although its impossible to know for sure what its downside is as you cant have 44.1 and 48 concurrently running for A/B of the exact same material. im just looking out for the future with my recordings by not limiting it [pardon the pun] to 44.1 from the onset.

 

from what i understand, there is a 2 channel uncompressed layer in DVD[V] at 24/96. there are MILLIONS of those players out in the world right now. unfortunately they arent in cars so... otherwise i would be releasing DVDv's with some eyecandy rolling along with the music.

[Reitzas]

My analogy to the audible difference between 44.1 and 48 is like the taste difference in putting 44.1 sugar cubes or 48 sugar cubes in your coffee. Sweet!

 

I use 44.1 almost exclusively.

 

As Alpha says, sample rate conversion is pretty good these days. The real pain in the ass is the waste of time and money in SRC of Mega-gig projects. When the tracks I just cut for a CD at 44.1 need to be remixed in a television special with other source material that was cut at 48, or if I am working on a project at 48 and the Artist and the band all want CD roughs at the end of a 16-hour day the last thing I want to think about is the time it takes for SRC. In time I know this wont be an issue but for now it is.

 

Even though I can hear the subtle differences as many of you have stated above, I just dont think its that critical of a component to making music. Bit rates, now thats a different story.

 

Dave Reitzas

[alphajerk]

"or if I am working on a project at 48 and the Artist and the band all want CD roughs at the end of a 16-hour day the last thing I want to think about is the time it takes for SRC."

 

hey dave, i got a real simple solution to this, just go analog out to a DAT [ha, and they thought it was dead] and make your index points and burn a cd from that. or for me [dont own a DAT] i just go analog out to my ADAT [not in use anymore] at 44.1 [i keep a tape in there just for this] and dump the tracks back in a new session... it takes a little longer but really no longer than bouncing to disk.

 

in reality however for those QUICKIES for bands... tapes the word of choice. im in no mood to burn a cd after 16 hours [but for enough money, i could do just about anything].

[Alon]

Tapes ? Who wants tapes anymore ?

 

EVERYBODY wants their own CD copy at the end of the day...

[Philip OKeefe]

Originally posted by Lee Flier:

Where the sampling rate really matters IMO is when applying DSP. EQ, dynamics and other DSP is more responsive and sounds better when it's applied to more audio. Now the benefit there between 44.1 and 48 may still be negligible, but at 96K you definitely notice the difference.<<

 

I agree with you here Lee.

 

>>I also have tracked everything on my AW4416 at 24 bit/44.1K so far, but I intend to re-cut a short piece at 48 just to see if the DSP is any more responsive. If it's that much better I can always record critical stuff at 48 and master it on my computer.

 

--Lee

 

Which, as you know, is what I always do. I DO have to lock at 48 KHz when using ADAT's with the AW (but can use 44.1 when locking my Dakota / Montana / Vegas / Logic stuff instead of ADATs).

 

I'd be interested to hear what you think after you track something at 48 KHz. One thing to consider though: Is the advantage of better DSP at the higher sample rate going to outweigh the disadvantage of doing a sample rate conversion? Which do you think will sound better? Well, try both and let me know which you prefer!

 

Just a thought.

 

 

Phil O'Keefe

Sound Sanctuary Recording

Riverside CA http://members.aol.com/ssanctuary/index.html

email: pokeefe777@msn.com

[Philip OKeefe]

Hey Alpha and Dave:

 

I use a discontinued Alesis AI-1 that does the sample rate conversion from 48 KHz to 44.1 in real time. you can convert from any input source to any output - any two ADAT tracks, S/PDIF, AES/EBU, etc.

 

I have software converters that might sound a bit better, but for doing a quickie at the end of the day, this sure beats waiting around for sample rate conversion software to run.

 

I'm really surprised that this box got discontinued. Garth Richardson turned me on to it several years ago while I was working on a album for Epitaph records with him, and I'm really glad he did. It's a very useful device, and if you can find one, you might want to pick one up.

 

 

Phil O'Keefe

Sound Sanctuary Recording

Riverside CA http://members.aol.com/ssanctuary/index.html

email: pokeefe777@msn.com

[alphajerk]

actually with the people dave works with i cant imagine an apogee psx100 not being there SRCing in real time to a CD burner. at least i think it does.

 

the gearboxing is pretty damn good from 48->44.1 now. i would be willing to bet one couldnt tell a difference between something started at 44.1 and src'd down to 44.1 in a blind test.

 

 

and yes, i give them cassette tapes, not CDr's. which with the people i work with works out great as all of them have cassette decks still in their cars and they love it being able to listen on the drive home.

[Max Ventura]

This might be a fantasy of mine, but I suspect that there are certain relationships between sampling rates and word lenghts that go well together and fit better with each other:

16 bit/44Khz, then 20/48, and 24/96. When working with this specs, work has always a "togetherness" in sound and definition that just "fits" better( than if yoyu go, for example, 24/44, or 16/96. But again, that could be a fantasy of mine.

[dansouth]

Originally posted by alphajerk:

SRC has gotten pretty damn good lately although its impossible to know for sure what its downside is as you cant have 44.1 and 48 concurrently running for A/B of the exact same material. im just looking out for the future with my recordings by not limiting it [pardon the pun] to 44.1 from the onset.

 

I guess I'm missing something. If sample rate conversion is "pretty damn good", how can the future of a recording be compromised by recording them at 44.1 instead of 48K?

 

My suspicion is that SRC is NOT that good, because it takes a lot of interpolation and extrapolation to move between 44.1 and 48K. This has to distort the program material - above and beyond the inherent distortion of quantization - at least as much as successive D/A and A/D conversions. If my concerns are founded, and if you want the best possible sound for future 24/96 fomats, then you are justified in recording everything at 48K.

 

However, the de facto distribution method today (for songs, not necessarily soundtracks) is 44.1K, and it will probably remain that way for years to come. Unless you expect a longer than average shelf life for your material, it seems prudent to work in 44.1 (or better yet 88.2) and avoid SRC as much as possible.

 

This is unfortunately one of those no clear win situations. Every choice demands a compromise. Let's all hope for a standard sampling frequency in the future, or at least even multiples of a standard.

 

In the final analysis, we might do well to contemplate how often any of this will matter. To people who listen on a walkman, car stereo, boom box, club P.A., or even an audiophile setup with a pricey tube/valve amplifier, there will be no audible difference between 44.1 and 48K. There won't even be a difference between 48 and 96K on these systems.

[soapbox]

Originally posted by dansouth@yahoo.com:

I've heard that DVD audio formats of 96 or 192K have been proposed, but so far they're nothing more than vaporware. A lot will change before these protocols are finalized, and they could eventually be killed by well-meaning bean counters.

 

Actually, DVD-Audio players hit the street a few months ago, as did the first DVD-Audio discs.

 

Originally posted by pokeefe777@msn.com:

I use a discontinued Alesis AI-1 that does the sample rate conversion from 48 KHz to 44.1 in real time. you can convert from any input source to any output - any two ADAT tracks, S/PDIF, AES/EBU, etc.

 

I have software converters that might sound a bit better, but for doing a quickie at the end of the day, this sure beats waiting around for sample rate conversion software to run ... It's a very useful device, and if you can find one, you might want to pick one up.

 

I agree. Its a great box for quick SRC. I used to work for a recording artist who got one when they first came out and we used it all the time.

 

BTW, I talked to several reps from Digidesign at their NAMM booth last week. They all said it would be some time before they support 96 kHz. Probably 2002. Otherwise, the upcoming ProTools 5.1 release will fully support all surround formats, as I'm sure most of you know.

 

This message has been edited by soapbox on 01-25-2001 at 06:19 PM

[soapbox]

Here's a link to a (6/00) Wired News article about the then imminent release of DVD-Audio, in case anyone is interested: http://www.wirednews.com/news/print/0,1294,36583,00.html

[Anderton]

>>Even though I can hear the subtle differences as many of you have stated above, I just dont think its that critical of a component to making music. Bit rates, now thats a different story.<<

 

I agree 100%. As to SRC, IMHO the real-time conversions are the ones you can hear. If you're willing to do it offline and specify the highest accuracy mode in your software, they sound pretty good. But when processing lots of megabytes, you might as well go out and have a sandwich while it crunches the numbers.

 

I decided a while ago to stay at 44.1 kHz just to make sure everyone worked in the same general way. Of course, the original ADAT threw a monkeywrench into that, as did DATs that only did 48 kHz...but overall, to the end listener, any difference is dwarfed by what's happening musically.

[soapbox]

Originally posted by Anderton:

overall, to the end listener, any difference is dwarfed by what's happening musically.

 

 

Absolutely. We gearheads often enjoy a good technical discussion, but of course the important thing is to nurture and capture a great performance in the studio.

 

Still, wed do well to remember that the listening public often chooses convenience over sonic quality. After all, cassette sales eclipsed LP sales before the CD became prominent; and right now, there are truckloads of college students whose modems are gulping down MP3s.

[Electrox]

Stick to the format you will be mixing in. Try not to sample convert. It is like losing a generation in the "old" days. Also, DVDs are a compression scheme. DVD-A isn't. Hopefully it will take off.

[alphajerk]

"I guess I'm missing something. If sample rate conversion is "pretty damn good", how can the future of a recording be compromised by recording them at 44.1 instead of 48K?"

 

because once you lose the top end @ 44.1 opposed to 48, its gone forever.

 

 

and about the compression of DVDv, yes in surround there is but there is also a 24/96 uncompressed STEREO layer that can be used.

[Lee Flier]

Phil wrote:

 

>>I'd be interested to hear what you think after you track something at 48 KHz. One thing to consider though: Is the advantage of better DSP at the higher sample rate going to outweigh the disadvantage of doing a sample rate conversion? Which do you think will sound better?<<

 

I have a feeling the 48 will still be better. This all depends on the quality of the sample rate conversion software of course. I have Cool Edit and I have done a few masters of DAT recordings where Cool Edit does the SRC and I really do not notice the degradation. It'll be interesting to see if I feel the same way about stuff that I transfer directly off the AW.

 

>>Well, try both and let me know which you prefer!<<

 

I will!

 

--Lee

[alphajerk]

i've SRC'd in cool edit vs. digital performers and there is absolutely no difference between those two programs [at the best resolution] but i have to basically let either go over night [well during the morning] as im sleeping.

[Alon]

I have to agree with those people who say they stick with 44.1, as more of a convenience issue.

 

I've always stayed away from 48, cause it only created hassles for me on any session I've done using 48.

 

I have always had Bad feelings about anything associated with 48.

 

Crappy consumer Dat machines, that only do 48, Crappy Adat's, etc.

 

Whenever I would transfer a bunch of Dats into ProTools, there's always that one crappy Dat that was done at 48, which messes with everything else.

 

You find yourself constantly looking at the Dat readout to check the sampling rate, when you switch Dat cassettes.

 

I just see 48 as a hassle.

 

If you need to incorporate SFX, or replace samples on your session, then these are most likely at 44.1

 

If your session is @48, then you'd have to convert SFX from a CD in order to use it.(In PT)

 

When it comes to doing a PRO session, speed is of the essence to me. People don't have the time to fuck around with unnecessary stuff.

 

44.1 sounds fine to me. (I can hear the difference between the 2 rates)

 

Now if I was working 88,96 or something like that, then I could see the point in SRC.

 

To be honest though, 44.1, 24 Bit is good enough for me. (For the time being...)

 

ALON

[soapbox]

This discussion has started up again at the Studio Tech forum:

 

http://www.musicplayer.com/ubb/Forum17/HTML/000658.html

[Anderton]

>>This discussion has started up again at the Studio Tech forum:<<

 

Like one of those pesky brush fires that just won't go out!!

[Guest]

man, it only takes a spark...

 

here's something that I have been wondering about-

 

I believe, maybe mistakenly, that technologies like L1 and UV22 work their magic with dither by redistributing energy into spectra where hearing is less acute. So what happens to that energy when we ignore "God's plan for our ears" and resample up to deliver in future formats...

 

what happens to a glob of signal when we ratchet up the Nyquist?

[dansouth]

I found one of Roger Nichols' recent comments interesting. 'Two Against Nature', Grammy winner for Best Engineered Album, was recorded using 16-bit converters and was mixed on an analog board to an Alesis Masterlink at 44.1KHz with 24-bit word length. Also, I seem to recall Mr. Reitzas saying that he uses 44.1 exclusively. I'm honored to be in such good company. http://www.musicplayer.com/ubb/smile.gif

[Jim Tavegia]

My 2 cents:

 

I do many location recordings with my Sony DAT, for convenience, and then come back to my studio and dump it into my computer for mastering and editing. I use an inexpensive Midiman Flying Calf 24 bit ADC in at 44.1 and playback (monitor) using the MAudio SuperDAC24/96. My computer has a DIO 24/48 card used as a digital I/O into Cool Edit. Midiman now has a DIO24/96 version PCI Card. I would think a Flying Cow 24/96 converter can't be too far behind.

 

I think Midman/MAudio offers the best bang for the buck out there for the home studio. They may not be Lucid/Apogee etc quality, but a huge improvement over most stock converters. If you have a marginal sounding CD player with digital out, try the SuperDAC. I do my best not to become a gear snob. All I care about is the sound my customer is happy with.

 

I tried the Delta 24/96 but could not get the noise floor down below -46db no mattter what I tried. With the DIO 24/48 I am below -80db. Good enough.

 

I will say this that with the inexpensive Delta even my wife liked the 24/96 wav files. She could tell a much smoother playback with more detail and air, a more three-dimentional sound stage. I hear that many like the sound of the PCI-based Card DeLuxe. It got a great review in Stereophile. It is too bad that this is not the CD RedBook standard, but this format is coming.

 

More importantaly, the 24/192 format for DVD-A is where we will most likely end up. Crystal and TI now both have 24/192 chip sets and I will bet that by early 2002 we will see the first ADC's and computer cards to match this new format.

 

Anyone who has heard 16 bit done with great ADC's and played back on a high quality CD Player ($1K and up) knows that there is great sound captured at 16/44.1, but you can't capture or play it back well with cheap converters and played back through lousy op amps. Unfortunately, that is what most consumers listen through with possibly high jitter brought into the mix as well. Any one who has heard SACD or HDCD knows the difference.

 

I can't hear the difference between 44.1 and 48K sample rates, but I sure can hear the difference between CD RedBook and 24/96. I also can hear the loss that comes from truncating down from 24 bit or even 16/48K to CD RedBook. I can hear a loss with Cool Edit or Red Rooster. That is why I stay at 16/44.1khz and live with what I start with. Maybe devices like the Lucid SRC are more transparent, but at near $2K I'd rather buy a better converter and stay at CD RedBook until 24/192 comes along. Then we are all going to need Gigbytes of RAM, not megs.

 

I have been experimenting recording at 24/96 into my computer, then playing back the audio through my Super DAC 24/96 and running that(analog) back into my DAT at CD RedBook through my 24 bit Flying Calf ADC at 44.1, eliminating truncation.

 

My first thought is that it is smoother with slightly more detail and 3-D effect, but I'm still doing more recording to prove this out. It is alot more work for what may be qestionable and marginal audio improvement.

 

It is very possible that a greater improvement could be made by upgrading to Grace qualtiy Mic Pre's and better Mic's than my AKG 3000B's or SM 81's. Each of us probably has weak links that are just as critical as converters and sample rates.

 

Monarchy Audio has a 24/96 capable jitter reduction box called the DIP that sell for about $250 that reclocks the digital stream and significanly reduces jitter. That will be my next purchase as I can use it during recording and/or playback.

 

Now, if we all could just win the "Big Game" just once, we could fix most of that. I know I could.

 

------------------

[Dylan]

Wasn't 48kHz put into effect by the "Man" only to stick it to consumers so they couldn't copy CD's digitally? The only time I'd bother using 48kHz is if I were working with a tape based digital format and was going to be mixing in the analog domain. Even then, I'd probably just stick with 44.1kHz. Sure, there are good software sample rate converters out there (i.e., Sound Forge), but most other software SRC's that I've heard definitely have audible artifacts (i.e., Wavelab). Often times I'll just resample the audio from D/A to A/D instead of SRC because of this reason. For practical purposes 44.1kHz is the best sample rate to use unless the project that your working on supports a different rate.

 

-Dylan