96K 24-bit?

Just a question from a newbie, working in HS2004...

If I'm typically burning my music to CD at 44K, what's the advantage to creating/recording music at higher smple/bit rates?

None really, so don't do it.

Seriously, you'll have a better MASTER recording, but then it has to be converted to 44.1 to burn to a CD. The downside is the files are going to be a LOT larger. It also is going to tax your computer more so unless you have a lot of horsepower, that's another reason to not do it.

Dave
Dave's Cakewalk FAQ
Computer: 2.4 GHz Core2Quad, 4GB/1.5TB+, Windows 7 Home Premium, Sonar 8.5.1Producer, Delta 44, (2)MIDISport 2x2, Mackie
Instruments: SL-880 Keyboard controller, Korg 05R/W, Korg N1R, Fender Stratocaster, Line6 Variax 300, Ovation Acoustic, Behringer GX212 amp.
Outboard: 1402-VLZ mixer, Digitech Vocalist WS EX, PODXTLive, and some other rack modules and stomp boxes.

To start, remember that to clearly sample a waveform, the wave must be sampled at least three times per cycle. So, when you run at 44.1kHz, your highest clearly-sampled frequency is only on the order of 14.7kHz. Since this is pretty close to the effective limits of most people's hearing range, as well as the limits of what most people's stereos can create, this is what CD's are sampled at.

However, if you've ever done anything like process digital photos, then you've probably discovered that effects work a lot better on higher-resolution pictures. If you want to post your picture to the web, you may reduce the resolution to the standard 96 pixels/inch (I think) for web pages. However, your camera may take a picture at something more on the order of 500 pixels/inch, or even higher. If you dither the picture down to web resolution before processing it and adding special effects, the results are much worse than if you process the picture first, then dither the results down to web resolution. A similar thing happens with sound. If you record, process, and mix at 96kHz, then mix the result down to 44.1kHz, the result is better than if you work entirely in 44.1kHz.

That being said, the improvement in sound quality is slight. Unless you are already pretty good at the countless other aspects of recording, mixing, and mastering, doing everything at a rate higher than 44.1 kHz is probably a waste of computer resources. And, since it takes much longer to do anything with 96kHz audio, you are also wasting your own time. Instead of working on your technical recording skills, you are sitting there waiting for your computer to process audio. Remember, once you really know what you're doing, you can always switch to 96kHz. But you are generally better served if you learn to walk before you try to run, skip, and jump.

As a final point for thought, higher frequencies are what define directionality of sound. The way these waves bounce around in the ear - and the resulting "distortion" or "echo pattern" - is what the brain uses to identify "location". The higher the pitch, the "clearer" this "distortion" is. Bats use incredibly high pitches, which allow them, in a very real sense, to "see" with their ears. In effect, they "see echos". At the high freqencies involved, the sound that reflects off of objects around them is as clear to them as the light reflecting off of them is to our eyes.

Humans can't compare to bats, but the human ear is capable of "feeling" sounds that are higher-pitched than the 15kHz (or at most 20kHz) that we can actually "hear". The brain parses this sensation in a similar way to the way it parses a sound. It *is* a sound, it's just too high for us to actually "hear" - i.e., the eardrum can't reverbrate that fast. However, the little hairs inside our ears still sense those waves, and we can tell what direction they are coming from. That's why some people can so clearly "hear" a 30kHz hum coming from a monitor, even though there is no physical way they can actually "hear" the sound. So, if you record at 96kHz (or 192kHz), the "direction" of sound can really blow you away. Of course, when you mix down to CD sample rates, you lose this, but you will have it in the future when the standard sample rate of the equivalent of CDs is higher than 44.1kHz.

That leads to one other reason that people record at 96kHz (or higher) - vanity. They feel that they are such awesome musicians that they must record in the highest-possible sampling rate for the "sake of posterity", even if their recording skills are nowhere near developed enough to warrant the high sample rate. But remember, if you don't have good recording and mixing skills, and high-quality (i.e. EXPENSIVE) mics and other equipment, you are wasting resources at high sample rates. Remember, a crystal-clear piece of garbage is still a piece of garbage, and the hum from a bad microphone sampled at 192kHz still sounds like the hum from a bad microphone.

ORIGINAL: daveny5

None really, so don't do it.

Seriously, you'll have a better MASTER recording, but then it has to be converted to 44.1 to burn to a CD. The downside is the files are going to be a LOT larger. It also is going to tax your computer more so unless you have a lot of horsepower, that's another reason to not do it.

Completely untrue.

24/96 gives much more information to begin with when recording. This means HS has to do less "guessing" or dithering when you start manipulating tracks with effects and mix down to your final stereo mix which could alter the original sound significantly if you recorded in 16/44.

I think that's what I said... it will give you a better master recording, but its still going to have to be converted to 44.1 to go on CD. Is that worth the extra overhead? I'd say for the typical home user, probably not. If you're working with orchestras and high end equipment, probably so, but then you wouldn't be using Cakewalk Home Studio.

Dave
Dave's Cakewalk FAQ
Computer: 2.4 GHz Core2Quad, 4GB/1.5TB+, Windows 7 Home Premium, Sonar 8.5.1Producer, Delta 44, (2)MIDISport 2x2, Mackie
Instruments: SL-880 Keyboard controller, Korg 05R/W, Korg N1R, Fender Stratocaster, Line6 Variax 300, Ovation Acoustic, Behringer GX212 amp.
Outboard: 1402-VLZ mixer, Digitech Vocalist WS EX, PODXTLive, and some other rack modules and stomp boxes.

Another point worth mentioning is that sample rate and bit depth are two seperate animals and affect different aspects of the recorded signal. You can record at 24 bit and take advantage of the increased dynamic range without using the higher sampling rates.

Of course your ability, gear, and the instruments you are recording have a big impact on the gains of working at 24 bit. Do some tests with acoustic instruments at 16 and 24 bit. In my setup, I can hear a significant difference when tracking my 12string acoustic guitar between 16 and 24 bit.

Do some searches of the techniques and sonar forums for bit depth and sample rate, there have been lengthy discussions...

[edit]Here's another discussion on sample rate and bit depth
http://www.recording.org/ftopict-26774.html[/edit]

ORIGINAL: daveny5

I think that's what I said... it will give you a better master recording, but its still going to have to be converted to 44.1 to go on CD. Is that worth the extra overhead? I'd say for the typical home user, probably not. If you're working with orchestras and high end equipment, probably so, but then you wouldn't be using Cakewalk Home Studio.

What kind of system are you using???? A Pentium 2 with a 40Gb HD????

This isn't video editing. 24/96 audio does not take up that much more space and isn't taxing the CPU like video does.

If your off the shelf PC from Best Buy can handle video, it sure as hell can handle 24/96 audio.

Gee, its great to have you back brzilian. I really missed your acerbic responses. At least is takes the heat off me.

I don't have an off-the-shelf Best Buy system.

I remember reading an article in Electonic Musician magazine a while back that said recording at 24/96 was not worth the overhead for typical home studio applications. If you want to use it, go ahead. YMMV!

Dave
Dave's Cakewalk FAQ
Computer: 2.4 GHz Core2Quad, 4GB/1.5TB+, Windows 7 Home Premium, Sonar 8.5.1Producer, Delta 44, (2)MIDISport 2x2, Mackie
Instruments: SL-880 Keyboard controller, Korg 05R/W, Korg N1R, Fender Stratocaster, Line6 Variax 300, Ovation Acoustic, Behringer GX212 amp.
Outboard: 1402-VLZ mixer, Digitech Vocalist WS EX, PODXTLive, and some other rack modules and stomp boxes.

24/96 audio does not take up that much more space and isn't taxing the CPU like video does.

If your off the shelf PC from Best Buy can handle video, it sure as hell can handle 24/96 audio.

I don't think the question should be "Can your PC handle it?", but "Are you gaining anything, and at what price? Is the price worth the gain?"

96/24 takes up roughly three times the space as 44.1/16, so you'll go through hard disc space and backup discs 3 times faster. It takes roughly 3 times as long to do any sort of processing. So, the amount of time you spend right now waiting for your computer, multiply it by 3 to get an idea of how much you'll be waiting if you go up. If you are currently having any trouble with dropouts, that problem will magnify, probably to the point of unusability. On the other hand, if you have a 4.2GHz dual-processor system, taking a 3x hit may be hardly noticeable.

That's the cost. What's the gain? For most beginners, there is absolutely none. There are so many details to recording, and so many things that are causing your recording to sound "not-quite professional", that 96/24 won't help you. As you progress, you will constantly be discovering ways to improve your sound, and all of them will have a much greater impact on the end project than going up to 96/24 will give you. So for a rank beginner, the question becomes "Should I take a 3x performance hit for no gain?", and I think the answer becomes plain.

Eventually, you will reach the point where things are sounding pretty good, and then going up WILL make a noticeable difference in your sound. Best thing to do at this point is just to go up to 44.1/24. You'll get a greater difference in sound in your end project simply by going up to 24bit than you will by going up to 96kHz sample rate. Eventually, when you feel you are really starting to know your way around, then go up to 96/24.

If you have an external MIDI keyboard, you can test things. I recommend using MIDI, because otherwise you will notice differences in the performance - it is nearly impossible to play exactly the same thing twice. Create a multi-track, multi-instrument MIDI song that you can play through your keyboard. Then create three different projects at 44.1/16, 44.1/24, and 96/24. Bounce the MIDI to Audio through your keyboard and soundcard in each project, and mix the results. Burn the final waves to CD. Listen to the results. To be fair, get someone else to play samples FOR you, so you don't know which one is playing, and see if you can RELIABLY tell the difference. If you can, then you'll know exactly how much you are gaining, and can decide if the cost is worth it.

I tend to record in 44.1/24; that seems best for my current needs/desires. Until late last year, I recorded in 44.1/16, and it was completely sufficient. And looking back at some of those projects, there are so many recording and mixing, well not exactly "errors", but things I could've done better, that I see for sure that it would have been a waste for those projects to be at 96/24. I was learning, so this is to be expected.

Good points, but the difference is really more like one third more disc space (24 is one third more than 16) and about twice the processor power (44.1 to 96 k). Scott Reams updated sonar test bears this out on the cpu load, at least for me. Simple math bears out the disc increase.

I'm running an A-64 so the load really is not the issue with me and I heaped myself up with hard disc. My habit is to master in 24/96 and dither to 16/44.1 when I need but with Homestudio this is no real issue unless you are doing lots of midi synths and have small hard drives.

But it was the same when I had an old P-II 400 that I was able to to do lots of audio recording on (6 tracks simulteneously at 20ish% cpu) with no problems and 24/96. It was in 98 rig and tweaked to the hilt. The only real time it bogged down was with soft synths so I routed my midi modules to audio tracks and that took care of that. I realize alot of people don't have the modules and need the soft synths, but sheesh, alot of computer can be had for relatively little money these days. (I stop to think about how my first used 8088 cost 600$, and then I had to buy the printer for 200 more, 20 years ago,... I paid less than twice that in real dollars today for what I have)

Not stray to far from the topic,... Agreed, you cannot decide what to do without testing and listening and understanding the limits of your equipment: then deciding what to do based on that but I do alot of straight audio recording and I would not even think of even starting a project unless in 24/96. I've developed the habits to make this work for me. I like my results. So do others.

but ultimately, the answer to the questions is so ideosyncratic that all I can say is try it and see.

Good points, but the difference is really more like one third more disc space (24 is one third more than 16) and about twice the processor power (44.1 to 96 k).

Well, not really - 24 is 50% (one-half) more than 16, and you also have to take into account roughly 2x more disc space for the higher sample rate. That would be right about the hit, though - the 24bit wouldn't make much difference there.

One more point occurs to me. It may seem like recording at 48kHz would be better than recording at 44.1kHz, because of the higher sample rate. But it doesn't really work out that way. The problem is that 48kHz has to be dithered to 44.1kHz. "Dithering" has a negative effect on sound. The 48kHz file has samples that are about .0208ms apart, while the 44.1kHz samples are about .0227. So, when the 48kHz file is converted to a 44.1kHz file, the first sample is unchanged. However, the 48kHz file does not contain a sample at .0227 seconds - it has one at .0208 seconds, and one at .0417 seconds, but not at .0227 seconds. So, in order to create the second sample for the 44.1kHz file, the computer needs to use the source samples from .0208 seconds and .0417 seconds to "guess" what the real value at .0227 seconds really was. This "guessing" introduces artifacts into the audio.

Note that the amount of guesswork required depends on the ratio between the two sample rates. It is very easy to convert an 88.2kHz signal into a 44.1kHz signal - simply throw out every-other sample. There is no "guesswork" at all, and the conversion is quite clean. Converting 96kHz into 44.1kHz involves some guesswork. However, that is not as bad as converting 48kHz into 44.1kHz. For 48kHz to 44.1kHz, there is a lot of guesswork involved - so much guesswork, in fact, that you pretty much defeat the whole purpose of recording at a rate faster than 44.1kHz. If you followed all this, you may guess that 88.2kHz would be even better than 96kHz if you want to burn to CD. This is true, but the 88.2kHz sample rate is not available on a lot of soundcards.

The math doesn't work that way... 24 bit is not 1/3 more than 16 bit because you're talking about binary.
16 bit gives you 65,536 (2 to the 16th power) different possible levels whereas 24 bit gives you 16,777,216 (2 to the 24th power) different levels. So a one minute recording at 16/44.1 will require 10.1MB of storage and the same recording at 24/96 requires 33.0MB. So actually 24/96K results in a file over three times larger than one in 16/44.1K.

For more info, go to:

http://www.tweakheadz.com/16_vs_24_bit_audio.htm

Dave
Dave's Cakewalk FAQ
Computer: 2.4 GHz Core2Quad, 4GB/1.5TB+, Windows 7 Home Premium, Sonar 8.5.1Producer, Delta 44, (2)MIDISport 2x2, Mackie
Instruments: SL-880 Keyboard controller, Korg 05R/W, Korg N1R, Fender Stratocaster, Line6 Variax 300, Ovation Acoustic, Behringer GX212 amp.
Outboard: 1402-VLZ mixer, Digitech Vocalist WS EX, PODXTLive, and some other rack modules and stomp boxes.

Boy did I miss that one. I did a acouple of tests and a 1 minute stereo file at 44.1/16 was 10.4 mb and the same at 99/24 was 32.7 mb.

Just the same it's not a real issue with me when I'm looking at two 160 gig drives. But it wasn't a real issue when I had only 80gig. I prefer to master at the highest resolution and dither down when I reduce to a playable wav for cd.

I'd still say try it and see. You'll know if you want it or not.

Dave and sinc,

You're both not seeing the big picture here. ALL entry level soundcards have switched to 24bit audio (including Sound Blasters).

24/96 is the entry point now for audio regardless of whether you're playing games or recording music (yes, games are now using 24/96 audio).

The Audiophile 2496 can now be bought for $99 as well as other entry level "prosumer". Heck, the AP2496 is probably going to be discontinued soon because M-Audio has released the Audiophile 192 at the orginal AP296 pricepoint when it came out ($189).

In a few years 24/192 audio will become the standard.

You're both not seeing the big picture here. ALL entry level soundcards have switched to 24bit audio (including Sound Blasters).

You've completely missed the point.

I've been trying to say that when you first start home recording, you might as well stay with 44.1/16 because it is lighter on resources, and you won't be using the greater sound quality. As your recording progresses, you will want to move up. What your soundcard supports has absolutely nothing to do with what I've been saying.

Sure, 5 years from now when we think of a GB the way we think of a MB today, people will switch. But that has absolutely nothing to do with TODAY.

You are also absolutely free to record at 96/24 if you want. If the hit on resources doesn't bother you, go ahead.