Best sampling frequency and D/A resolution

  • Thread starter Thread starter Arjen
  • Start date Start date
A

Arjen

New member
Hi,
I want to record audio with a 24/96 soundcard (haven´t got it yet). I ´ve already recorded with a SBlive (16 bits and 44kHz) and since I´m going to buy a new soundcard my question is:
Which is the difference between recording at 96kHz and 44 if I´m going to save my work as AudioCD (44kHz)?

If the difference is apreciable which computer is needed for 30 tracks recording at 24 bits and 96 khz?
Also any soundcard recomendation would be appreciated.
Thanks
 
What I'm about to say is more of an opinion than anything. I say that because this topic is of much debate.

I personally think that bit resolution is more important that sample rate, although even jumping from 16bit to 24bit is subtle. There's no question that 24bit converters have the potential to be much better than 16bit converters (especially when you see your noise floor down under -96db!), but whether you actually record using 24 or 16 bit resolution with those converters might not make as much of a difference as you might think.

But anyways, the reason I dislike 96khz recording is that it more than doubles the amount of data your system has to work with for little audible gain (on a consumer system in a home studio etc). Any improvement in fidelity can surely be lost by sub-par downsampling technique. Also, a lot of plugins will not work at 96k, even some of the more expensive ones. Then you have to deal with the fact that you can't work with multiple sample rates in a single project without first converting everything to one sample rate. So if you're using fruity loops for some stuff, for example, you can leave the tracks at 16bit even if your project is 24bit, but you'll have to convert them to 96khz if that's what the rest of your audio is at. It is possible that some software will automatically convert sample rates in realtime, but are you going to trust what you're hearing?

In theory, higher sample rate and greater bit resolution are good ideas. The more data a reverb plugin has to work with, for example, the better it should sound. The higher the bit resolution, the smoother your edits can be. And so on....

If you're not a professional with a professional setup, then do not use 96khz unless you've proven to yourself that you can hear the difference. It will drastically reduce the capabilities of your DAW. In fact, don't use 48khz either unless you can prove to yourself that the difference is noticable.

Personally, I record at 24/44...it's very easy to work with and sounds as good as I expect it to.

Now don't turn my opinion here against me. I'm not saying that a 24/96 soundcard isn't important. I'm saying that the *numbers* are less important than the sound of the card itself. If you jump up to something like a Delta from a soundblaster, you will immediately notice a drastic improvement in sound quality, EVEN if you record at 16/44! The converters are better...the card is better...the sound is better. That is what is important. Don't get too hung up on magazine hype!

Slackmaster 2000
 
Unless you have a pretty stellar signal chain I wouldn't worry too much about 96khz. 24bit 44.1 khz is still pretty standard (and damn good).
 
i've noticed a more open sound with 48khz also using a spectral analyzer i see their are more harmonics between the 4khz and 20khz area then when i record @ 44.1 but this is with a delta 66(no omni studio) so other cards may vary

another advantage of higher bit and sample rate is the processing will be much 'cleaner' errors in the math will be alot less audible
 
Thanks to all of you for helping me in my recording adventure.
 
When all is said and done your 96KHz or 48KHz file at 24 bits needs to be mixed down to 44.1K-16 bits to burn. After you master it with multiband compression and multiband limiters, there may be a slight advantage on the high end, but sometimes not even audibly better. Cymbals may be slightly brighter but not much else.

I have heard others say that when you record at 44.1/24 you get warmer mids than at the higher resolutions. This too can be desirable for some things.

I say 44.1 - 24 bit. Usually excellent results with much lower computer resource usage as others have already stated.

To answer your last question. I am running a P4 - 2.26 overclocked to 2.4 and getting up to 50 tracks with EQ and compression on about 1/3 of them. Reverb on about 5. No dropouts. Very smooth. 2 or 3 Synths and sound modules are also running on my largest project.

I am not sure you could run 30 tracks on a PIII running at 1Ghz without dropouts. Prior I had a PIII running at 811Ghz and was getting dropouts on around 15 tracks.

Minimum, I'll go out on a limb and say a P4 1.6 - 1.8 would do.

Soundcard reccomendation would be any of the M-Audio cards. I use the 2496 and it is crystal clear. Mine is a writing oriented studio and, if you are recording live, you would not want its limited 2 ins and 2 outs. If you need a lot of inputs one of the other M-audio cards would be better. I have also heard good things about Terratec cards.
 
I like 24bit/88.2kHz. The high sample rate gives a slightly more warm and open sound to processed signals, plus the math is very simple when resampling to 44.1kHz.

barefoot
 
I started using 24/96 resolution a few months ago. I feel like my mixes are sounding much better - but there are other factors contributing to this. I don't think you need a super expensive studio for thisto make a difference. The quality of your FX is equally important to that of your mics/pres ect. In fact, the difference between semi proffesional gear (like a $400 pre-amp and a $600 mic) and the real good stuff is much more subtle than the difference between the OK stuff and the total crap...I am rambling.

Hi res audio is definitely more taxing on the system - and it gobbles HD space - it's pretty amazing to see your available space go down at something like 2MBs a sec (recording 8 tracks).

One interesting thing though - at 88.2K and 96K sample rates, you can pick up higher frequency sounds (perhaps as high as 40KHz). Now, what that means if mics aren't sensitive in that range, and neither are speakers - I don't know.
 
One interesting thing though - at 88.2K and 96K sample rates, you can pick up higher frequency sounds (perhaps as high as 40KHz). Now, what that means if mics aren't sensitive in that range, and neither are speakers - I don't know.
Click to expand...

y does it matter if your speakers can properly duplicate a sound at 40khz when you can barely hear over 12khz?
 
y does it matter if your speakers can properly duplicate a sound at 40khz when you can barely hear over 12khz?
Click to expand...

Some people can hear as high as 22khz. But the main argument I've heard is that the higher frequency sounds that you can't hear interact with the lower frequency ones which you can.
 
Just because you can't hear a frequency present in your signal doesn't mean that it doesn't exist and isn't having an impact on what you hear, and the processing you do to the signal.

Yes, any complex waveform can be broken down into thousands of individual sine waves that when added, produce your original waveform. This does not mean that you should necessarily think of your signal as a collection of individual frequencies, however. That's too simplistic.

That said, I still can't hear enough benefit at 96khz to use it. Take a project that you've recorded at 24/96 and convert all the wave files to 24/48 and see if you can hear a difference between the two mixes.

Slackmaster 2000
 
No, No, No.

The benefits of higher sample rates have nothing to with the audibility of ultrasonic frequencies. :rolleyes:

The benefits are mainly twofold:

1. Low-pass filters are employed above 20kHz to prevent "alias" signals from corrupting the audible range. Aliases are ultrasonic signals that sort of get transposed down into the audible range due to the technicalities of sampling. The effects of these sharp low-pass filters bleed down into the audible range and cause phase and transient distortion. Using higher sampling rates allows the use of less drastic low-pass filtering to prevent aliases which in turn cause less phase and transient distortion in the audible range.

2. The second reason is a little more esoteric, but probably the most important. The basic point is that higher sampling rates provide more "precision" for certain types processing. Even though your hearing can’t reach out that far, some aspects of the mathematics involved in processing require a wider bandwidth in order to avoid errors in the audible range.

barefoot
 
You must log in or register to reply here.
Back
Top