Sample Rate: 44.1 kHz, 48 kHz, or higher?

[lonemale]

I'm wondering what In/Out Sample Rate to use in the Audio tab in the preferences in Live 6?
I have a fast computer and a RME Fireface soundcard. Should I go with 48, or just stick to 44.1.... or is 96 where I should be?

I use 24-bit samples, and do most of my mixdowns in Live - not using any other DAW.

Suggestions, explanations, ridicule, "what does it all mean..."? Bring em on.

[Tone Deft]

take a look at what most of your gear uses and stick with that. CD quality is only 16 bit 44.1kHz. Maybe the only issue is conversion from one sample rate to another. In the end it doesn't matter THAT much, just keep it simple.

[ChiDJ]

Stick with 24 bit and 44.1.

[dj superflat]

there's a fairly good rational for going out to 80 (higher rate, yet easier math (divide by 2) to get down to what you ultimately use. with disk storage where it is, i don't see memory use as a real concern any more.

[dancerchris]

Nyquist theory shows that sample rates above 44 do nothing for the quality of recording for human audible frequencies. However:

high sample rates (88.2 and 96) decrease the latency of your system, some also claim that the processing of the signal internal to the DAW is done with less error roundoff error.

More important is the bit depth (16 vs. 24) and that is a no brainer, try to always sample at 24 rather than 16. You get much more headroom above the noise floor for signal processing.

The down side: high sample rates and bit depths increase the processing and storage requirements of your equipment.

Remember that: Perfectly recorded garbage is worth less than poorly recorded gold.

[ChiDJ]

Nyquist theory shows that sample rates above 44 do nothing for the quality of recording for human audible frequencies. However:

high sample rates (88.2 and 96) decrease the latency of your system, some also claim that the processing of the signal internal to the DAW is done with less error roundoff error.

More important is the bit depth (16 vs. 24) and that is a no brainer, try to always sample at 24 rather than 16. You get much more headroom above the noise floor for signal processing.

The down side: high sample rates and bit depths increase the processing and storage requirements of your equipment.

Remember that: Perfectly recorded garbage is worth less than poorly recorded gold.

Very well said.

[compositeone]

Amon Tobin said in an interview for knowledge mag that it's pointless to to go higher than 44.1.

Right or wrong if its good enough for Amon its good enough for me.

[zstowasser]

well said mr. chris! I've heard what you said from a variety of sources so I know you're on point. I've recently swtiched to 24bit as well but keeping at 44.1 sample rate.

I've also heard that since almost all music players and consumer soundcards are based around 44.1, changing to 48 or 96 would cause odd mathmatical rounding, and choosing 88.2 or another clean multiple of 44.1 would be a wiser choice, if you chose to change the sample rate

However a question that I haven't heard answered is what about DVD / HD-DVD / Blue-Ray and the future of high definition audio, since they are at 48, 96 and 192.. what are the reasons to stay at 44.1 rather than switch to 48 to match with the high-definition standards?

[ethios4]

Nyquist theory shows that sample rates above 44 do nothing for the quality of recording for human audible frequencies.

Another factor to consider : to prevent Nyquist aliasing, a low-pass filter has to be applied to the signal to remove the frequencies above the nyquist cutoff. No filter is perfect, so this process will introduce unwanted distortions in the area near the cutoff. If you are working in a higher sample rate, these distortions will be out of the range of human hearing and you get cleaner highs.

[mechmusic]

[quote="dancerchris"]Nyquist theory shows that sample rates above 44 do nothing for the quality of recording for human audible frequencies. [quote]

This theory goes out the door when you have a software which is specialized in time stretching and pitch shifting of audio files! If you slow down the tempo to half speed while keeping the pitch constant, your sound quality will be better with higher sampling rates. Similarly, if you pitch shift samples to extreme lower frequencies, your sound quality will be better with higher sampling rates.

Try it out on a few samples. If you're not pitch shifting and time stretching audio, then sure use 44.1, but if you are then consider 96.

[dancerchris]

Nyquist theory is a mathematical fact. I mentioned in my original post that the roundoff error is smaller with higher frequencies and this extends to processing errors for something like timestretching (warping).

I used to be concerned about the antialiasing of the hardware and suffice it to say unless you know how the particular hardware your using adresses the antialiasing (oversampling or sophisticated "brick wall" filter) going to 96 may be redundant i.e. oversampling. Additionally your hardware performance may roll off with increased sample rate. Most manufacturers won't tell you what their antialiasing method is (in fact they quite flatly tell you it's proprietary, but many internally oversample).

As for rounding errors during up or down sampling this is just a myth as there are only whole numbers used in the process.

96 and 48 are very usefull for the video industry.

By the way, just for giggles, I use 96/24 only because I get really low latencies on my setup. Lots of Pros use 44.1 (even 16 bit) and do great.

Remember that: Perfectly recorded garbage is worth less than poorly recorded gold.

[ethios4]

Remember that: Perfectly recorded garbage is worth less than poorly recorded gold.

lol....and thats pretty much the bottom line.

[mechmusic]

Mathematical fact is that if you record a 120 bpm loop at 44.1 khz, then you warp it to fit a song tempo of 90 bpm, you aren't listening to a 44.1 khz sample anymore! it's now a 33 khz sample.

So, I'm not talking about processing errors or round off errors.

Think about it this way, if you originally record 44,100 samples for one second, but you slow down the playback so it takes 1.33 seconds to play back, you still have exactly 44,100 samples /1.33 seconds which equals 33158 samples/second or 33khz. You have in effect reduced the sampling rate.

Now, if you started out with a 96 khz sample you could reduce the tempo by half, and still have a 48 khz sample. Your sound quality will be better maintained.

I could explain how pitch shifting to lower pitches also changes the sample rate in a similar way, but I hope my explaination above is understandable.

[difference]

dancerchris is 100% spot on.

Worth noting though that the cinema industry tends to use 48 or 96 Khz.

I have read somewhere that it is a "better" conversion from 88.2 to 48 than it is from 96 to 44.1, so if you are making something that could be used for both purposes 88.2 khz is the way forwards.

[D K]

i'd flush opinion and do some tests, and use whatever personally pleases your ears the most.