question about sample rates

Jim Phelps · Post by **Jim Phelps** » 5 Dec 2006 1:23 am

I've been using 44.1 Khz sampling rate and 16 bit file bit depth on my home recordings. What is the normal "studio-quality" setting these days? I know theoretically that faster sampling and more bits is better but is there a practical limit? My pull-down menus show that I can select up to 96 Khz sampling and 24 bits. Should I use it? If I record a track for another studio, I need to use whatever settings they're using on the project, is that right?[This message was edited by Jim Phelps on 05 December 2006 at 01:51 AM.]

mtulbert · Post by **mtulbert** » 5 Dec 2006 5:54 am

Good question there.

The rates keep going up and up but the final product is still done in 44.1 16bit. A higher res recording will have to dithered at some time to get it done to 44.1 and 16bit. I guess you would have to check with the studio that you are going to share files with to determine what they need.

Also, remember that the size of the files increase dramatically as the bit depth and resolution are increased.

Regards,

Mark T.

PS Everything I do in my little studio is at 44.1 and 16bit. Seems to sound great here and avoids alot of confusion as well.

M

Greg Cutshaw · Post by **Greg Cutshaw** » 5 Dec 2006 6:13 am

One of the reasons I bought the Yamaha AW-1600 workstation was that it offered 24 bit recording. Everything sounded absolutely stellar in 24 bits so I thought that was important. Well, my files sizes with all the track .wav files were approaching 1 GB even after optimizing. I tried a few songs at 16 bits, 44.1 khz and to my ears there is absolutely no difference in the quality. Even the highest steel parts and the drum cymbals are sparkling clear with no hash. In my case most of these get converted to .wma files at 192khz anyway and they still have great fidelity.

So I record everything in 16 bits now and my file sizes are greatly reduced. This saves a lot of file space because I archive to 2 independent disks and it also speeds up the USB bus transfer time.

Greg

Steinar Gregertsen · Post by **Steinar Gregertsen** » 5 Dec 2006 6:59 am

You won't notice much of a difference from 16 to 24 bits unless you're recording very 'sensitive' acoustic material.

The big advantage with recording in 24bits is that with heavy editing/processing you lose some bits, a 24bit recording will give you more headroom (please don't ask me to explain the tech details, I don't know how, but this is what everybody is saying on the rec forums so I assume it's correct).

You can compare it to using a steel guitar amp with lots of headroom before the sound starts to get 'dirty'....

Personally I record in 24/44.1 and am happy with that solution..

Steinar

------------------
"Play to express, not to impress"
www.gregertsen.com
Southern Moon Northern Lights

[This message was edited by Steinar Gregertsen on 05 December 2006 at 07:50 AM.]

Hal Schmid · Post by **Hal Schmid** » 5 Dec 2006 9:10 am

I record at 24/44.1 as well.

Steve Stallings · Post by **Steve Stallings** » 5 Dec 2006 9:47 am

24/44.1

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

Steve Stallings
The Songs

Jim Phelps · Post by **Jim Phelps** » 5 Dec 2006 11:47 pm

Thanks, everybody.

Frederic Mabrut · Post by **Frederic Mabrut** » 6 Dec 2006 12:09 am

I prefer 16bits 44Khz for demos and 24Bits 48khz for pro recording.

Sampling rate = Fs 2.2 x F
So if you want record the wholeness of what a human ear can ear (up to 20Khz) you must record at 2.2 x 20 =44Khz
At 48 and 96Khz you got a more precise and nearly perfect render for the very high notes and sounds.
The number of Bits defines the resolution of the sample and gives the maximal dynamic.
So you got 90Db in 16 bits and 138 db in 24 bits. At higher resolution you also have a much more precise possibility to adjust things such as plugins, etc..
Hope this will help

Tony Prior · Post by **Tony Prior** » 6 Dec 2006 3:28 am

for several years I recorded with my Boss BR-8 which is 16 bit..It was really a fine tool and produced excellent results( to my ears) . I wrote EQ and effects patches specific for my guitars. I used to get all these comments from Steel players that my Push Pull had a great tone. Problem was my Push Pull was a pre BCT Carter !.

I upgraded to a Yamaha AW2816 which records in 16 and 24 bit. For the first few months I recorded in 24 bit, but I was never quite satisfied, it seemd tight, stiff, something..so I backed down to 16 bit..and I cannot put my finger on it, maybe it's my aged ears..but I prefer 16 bit to 24 bit...

I'll let the 24 bit experts tell me whats wrong with me, but for now I reside in the 16 bit domain !..

Ray Minich · Post by **Ray Minich** » 6 Dec 2006 6:18 am

Do the DAC's have the precision to slew at 24 bit rates? Or are we precision on the ADC's, digitally correct in our effects, and then sloppy on the DAC's?

Jack Stoner · Post by **Jack Stoner** » 6 Dec 2006 11:10 am

Then after you record at the higher rates, when you make a CD, it is recorded at 16 bits (all commercial CD's are 16 bit).

Bryan Daste · Post by **Bryan Daste** » 6 Dec 2006 1:58 pm

Basically, as I've understood it, recording and mixing at 24-bit resolution will give you improved clarity and resolution, even when it gets put onto a 16-bit CD; if your DAW has a dither option, use it. Dither is what allows you to retain that resolution.

Here's some more info on dither: http://www.pcrecording.com/dither.htm

As far as sample rates, there's a lot of debate on this issue. One thing people seem to agree on is that you should avoid 'gearboxing' - going from 48kHz to 44.1 kHz by re-sampling the audio is 'gearboxing.' If I know my audio is eventually going to a CD, I will record at 24 bits and make sure my sample rate is a multiple of the CD standard - either 44.1 kHz, 88.2 kHz, etc.

For a comprehensive look at how the nuts and bolts of digital audio work, check out this book by my former U of Miami professor:
http://www.amazon.com/Principles-Digital-Audio-Ken-Pohlmann/dp/0071348190

Bob Martin · Post by **Bob Martin** » 6 Dec 2006 10:37 pm

I use 24 bit 44.1 but one quick comment about dithering. The higher quality that you use to dither the more CPU cycles it uses and I understand we would all be better off to turn dithering off while we are tracking but make sure you turn it back on when you start to mix.

I didn't one time and everything I mixed down with no dithering came out very harsh and almost unusable. Well luckily it was just one song and the problem was you don't hear the dithering process while you are mixing it only happens when you export as a final mix hence the mixup for me.

Now I use Sonar and it might not be that way on hardware HD recorders you might have to dither all the time but since I learned the little dithering trick in Sonar I'm saving lot's of CPU cycles and things are rendering much faster. In Sonar there are at least 5 or 6 dithering choices.

Play with it and see it might not work this way even on your software recorder.

Bob

Steinar Gregertsen · Post by **Steinar Gregertsen** » 6 Dec 2006 11:01 pm

I always mix down to 24 bit or 32 bit 'float' and leave it at that until I either send it off for mastering or master it myself.
Dithering must always be the very last thing you do to a mix, 'multidithering' is not to be recommended. I only dither to 16 bit after I have done all mastering edits, including fades, then I save and never touch it again.
If you have a wave editor with a "bit meter", like WaveLab, you'll notice that any changes to the wave file influences the bit rate.

Steinar

------------------
"Play to express, not to impress"
www.gregertsen.com
Southern Moon Northern Lights

Jim Phelps · Post by **Jim Phelps** » 12 Dec 2006 12:29 am

Thanks again for all the help, everybody.

Brad Sarno · Post by **Brad Sarno** » 12 Dec 2006 6:01 am

Yea, keep it 24bits while you record and mix. 44.1kHz is fine so you can just keep that set from beginning to end. The 24 bit factor will help keep your "resolution" higher. This will especially help when mixing your tracks together. The math is MUCH higher compared to 16bits. Try to keep the whole project 24bits until the very, very end of the "mastering" stage, and then it can get dumbed down to 16bits to fit the CD standard.

To get technical, your "bits" has to do with volume or amplitude, NOT frequency response. An audio waveform has two axis. The horizontal axis is time or frequency. The vertical axis is volume or amplitude. The "bits" determines how many possible steps there are to determine amplitude.

A bit depth of 16bits gives you 65,536 voltage steps.

A bit depth of 24bits gives you 16,777,215 possible voltages to define volume or amplitude.

You can see there's a huge difference. Sonically, this isn't always an obvious thing if your music is all pretty loud and doesn't have soft passages or long reverb trails. But to a good ear, even on loud music, you can tell the difference. The 24bit factor becomes even more significant if you're mixing many tracks together.

Brad

Jim Phelps · Post by **Jim Phelps** » 12 Dec 2006 6:50 am

Great info, thanks Brad.

John Macy · Post by **John Macy** » 12 Dec 2006 7:35 am

Good info here, and Brad is right on the money, buuutttt...

Sometimes you have to use your ears to find what you want. I think a well designed and clocked 44.1/16 bit system can run circles around a average ( or in many cases piss-poor) 24 bit/96K system. I have witnessed in person many, many tests with killer engineers where they have picked 44.1/16 bit stuff as the 96K stuff and thought the 96K stuff was the 44.1/16. Of course, we are talking about the PARIS system here, which is extremely well designed and externally clocked. I have also watched Brian Tankersley mix 100+ track counts in PARIS, with tons of plugins, 16/44.1 with no summing problems at all. My average mixes are between 48 and 64 tracks and sum just fine at mixdown.

I recently had a project come in for a mix that was done 88.2/24 Protools and the client was very unhappy with the mix.. I downsampled it to 44.1/24, and for grins 44.1/16. Remixed one song both ways. The client actually chose the 44.1/16....

When I track anything delicate and acoustic, I do use 24 bit. When I track anything in ProTools, I use 24 bit. When I track in PARIS for anything band oriented, I generally use 44.1/16 'cause to my ears (and the ears of my clients and my mastering engineer) it just sounds bigger and better.

So I would say experiment with your system and see what sounds good to you. A better word clock will often make a way bigger difference than a higher sample rate or bit depth.

I think there is way too much information (and mis-information) out there in the trade magazines and forums that people often listen too much with their eyes and not their ears. I tracked an record for an artist last year that was really good, the artist was thrilled---I got a call from her manager and her label telling me it was the best sounding project that she had ever done. Well, she decided to do some keyboard overdubs at a small studio, and needed some .wav file--to I put them on a drive for them to work with. I got a call the next day asking for the 24 bit files (mind you, this is a keyboard overdub session). I told them they did not exist, cause I tracked it at 16/44.1. She went ballistic on me--didn't matter that up to that moment everyone on her team, including her, thought it sounded amazing. Turns out this keyboard guy had a basic ProTools 001 system and had never made a record, but showed her some magazine articles saying 24 bit was "better". He even told her that using a PC instead of a Mac was bad for her sound quality. Anyway, I passed on finishing the project...

"So what do you think of that take?" "I dunno, looks like it sounds good..."

David L. Donald · Post by **David L. Donald** » 15 Dec 2006 11:52 pm

What Brad and John say.

I am doing the current 'flow' project in
48khz 24 bit,
on my 'not updated because of french divorce' personal system.

For light acoustic stuff I would also go 96khz 24 bit.
If I had it right now.
I don't have it for the reasons of not wanting to get a cheap cheesy system.
A waste of sheckles. 48/24 is sufficing.

James Quackenbush · Post by **James Quackenbush** » 16 Dec 2006 10:39 am

John M,
You took the words right out of my mouth !!..The comparison I was going to make would have been using a RADAR setup , but your comparison running Paris is every bit as convincing ..Quality of digital convertors , and the quality of the dithering tool has a lot to do with the sound ... Happy Holidays...Jim

Brad,
Have you even tried recording at 88.2 to see if the dithering down process is cleaner due to the 44.1 being exactly 1/2 of 88.2 ??.....I have been meaning to try this , but would like to hear from you on this ...Thanks, and Happy Holidays...Jim

Ray Minich · Post by **Ray Minich** » 8 Jan 2007 10:15 am

A bit depth of 24bits gives you 16,777,215 possible voltages to define volume or amplitude.

Do the A to D's and D to A's really have the resolution to track that minutely? Or is the resolution lost to the reactance monsters?