Best Sample Rate

well considering that your using such an old computer im not sure working at a higher sample rate will work for you. the fact that your using Audio Instruments will even make it harder for you to operate efficiently at a higher sample rate that 44100. if you upgrade to a faster computer,, much faster computer you should very much consider working at a higher sample rate as it will yield much more natural sounding results when creating music digitally. i produce music in logic at 96k and will never go back to 44100.. when you look at the science of how audio is created and or processed digitally you will find that higher sample rates are the best choice for producing music in the digital realm...

24-bit, 44.1 KHz.

when you look at the science of how audio is created and or processed
digitally you will find that higher sample rates are the best choice for
producing music in the digital realm...

When you look at the science of it, you'll see that 44.1KHz is perfectly capable of reconstructing the source waveform within the bandwidth of human hearing competely accurately.

Science wise, there is really only one main advantage of going to higher sample rates, and that's to require less steep anti-aliasing filters - some softsynths therefore sound better in terms of aliasing and produce a cleaner output.

Art wise, some people prefer 96KHz, especially people who really have the studio, gear, accurate monitoring environment and the ears to really benefit from it. But 96KHz is mostly a pointless exercise in my view if you are just recording audio.

Halving your entire system performance to go 96KHz *for me* is mostly not worth it - the gains do not balance out the losses.

It all ends up as mp3's these days anyway... headbang

I agree with BeeJay (and noeqplease) in that sample rate, in itself is not such a big question you should be asking yourself. Bit depth...... that's a different matter.

In a digital, multitrack situation you should record your audio at 24 bit (as you state in your post). It gives you greater headroom and dynamic range and handles summing into stereo way better than 16 bit. Also, you can track at much lower levels and multitracking in DAW's really benefit from keeping those levels down.

I always choose sample rate depending on the project. If it is an album and is going to mastering, it is 48k @ 24 bit.

If it's destined for CD and NOT going to mastering then 44.1 @ 24 bit. Sometimes stepping down to 44.1k from a higher sample rate can do more harm than just recording at 44.1 to start with. If you're sending to a pro mastering house they have way better tools for sample rate conversion and like to have the full bandwidth version to master from.

DVD or film, 48k @ 24bit.

Hope this helps some.

In your case,

24 bit, 44.1 kHz.

Cheers

blah blah blah.... such a ridiculous debate that will some day come to a rest,,, like **** the world will be using 44100 in the years to come.... have fun in low res land.......

What an elegant and well-reasoned argument. You've convinced me!

I'm going to run everything at 192KHz now. In fact, I'm going to upsample all my old grungy 8-bit samples to 192KHz to take full advantage of high-res fidelity.

Whee!

hehe,, at the end of the day there is no argument,,, when using higher sample rates you will achieve better results...

would you rather watch Standard Definition TV or HD TV?...

when using higher sample rates you will achieve better results...

Not if I don't have the resources to practically achieve what I need to at higher sample rates 😉

would you rather watch Standard Definition TV or HD TV?...

Actually, to be nit picky, that analogy doesn't quite work as increasing the resolution of the picture (from standard def to high def) is analagous of increasing the bit depth of audio.

To have a matching analogy of high sample rates would be like watching TV at 100 frames per second, rather than 25.

In any case, I'm of the opinion that the practical energy above the commonly accepted human hearing is minimal, and therefore devoting 75% of my resources on something that I can't hear doesn't make sense, except in the one case I already mentioned where the practicalities of high sample rates on aliasing performance does extend into the conventional frequency range.

44.1KHz is capable of perfectly reconstructing any analog waveform we can hear, and that's good enough for me. And at the end of the day, the final sample rate is always going to come down to the regular ranges, so any benefits of preserving this elusive high-high-end are going to be lost at the end of the chain anyway (not many consumers are listening at more than 44.1KHz).

Anyway, I don't want to turn this thread into a sample rate war, so I'll leave it there and feel my opinions are adequately expressed... 😉

Bee Jay wrote:
Anyway, I don't want to turn this thread into a sample rate war, so I'll leave it there and feel my opinions are adequately expressed... 😉

Who doesn't love a good sample rate war ? 🙂 Naw, but just because I can't resist being nit-picky myself...

In any case, I'm of the opinion that the practical energy above the commonly accepted human hearing is minimal, and therefore devoting 75% of my resources on something that I can't hear doesn't make sense, except in the one case I already mentioned where the practicalities of high sample rates on aliasing performance does extend into the conventional frequency range.

Higher sampling rates allow more gentle lowpass filtering, as you mentioned. That's true for all audio, not just for softsynths, of course. Frequency response above 22kHz may have no practical significance, but in the current world of imperfect filters, it seems beyond dispute that 44.1kHz is a borderline sampling rate--it places the LPF close enough to the range of verifiable human hearing that aliasing and phase artifacts are significant concerns in the design of D/A converters. That's not to say that 96kHz is "worth it" in anyone's particular situation, but there is some legitimate math on the side of higher sampling rates. The very existence of oversampling is testament to that...

44.1KHz is capable of perfectly reconstructing any analog waveform we can hear, and that's good enough for me.

Not technically true, and as a tangent, this is one of the most common misconceptions about sampling theory. Nyquist-Shannon proves complete reconstruction, but with the assumption that every sample is perfect, which it is not for digital audio. Each audio sample is subject to quantization error, and that error is reduced but not eliminated by moving from 16-bit to 24-bit samples.

when using higher sample rates you will achieve better results...

Not if I don't have the resources to practically achieve what I need to at higher sample rates 😉

Ah... the best possible argument for sticking to 44.1 ! 🙂

Seriously, thanks for the explanation, Bee Jay... plenty of valid points. Bottom line advice for everyone, I think, is: if your gear is capable of higher sampling rates, TRY IT OUT. If you hear improvement and can afford the performance hit, use it. Even if the math is accurate, this question cannot be answered with math. Sampling theory deals with ideals that do not exist with current audio gear... the closer you look at the details of your converters, the more the basic sampling math goes out the window... all converters exhibit measurable jitter, imperfect LPFs, approximations of infinite sinc functions, etc., and with all due respect to its beauty and elegance, Nyquist-Shannon addresses none of these things. Ultimately, once again, we're left with the uncomfortable position of actually having to listen and decide for ourselves 🙂

James

thank you Mr Nash,,,, now can you also comment on plug-ins being used at higher sample rates.

would a higher sample rate force a plug-in to operate at a higher precision,? I.E. the attack and release of a dynamic controller and or the amplification or reduction of frequencies via an Equalizer?.

it seems beyond dispute that 44.1kHz is a borderline sampling rate--it places the LPF close enough to the range of verifiable human hearing that aliasing and phase artifacts are significant concerns in the design of D/A converters.

Yes, very true. It seems that sample rates of about 50-60KHz would have been a better practical standard rate for this reason. Once you go above this, the law of diminishing returns applies, and you have to factor in the convertor performance. Ideally, 44 is too low, and 96 is too high.

. Bottom line advice for everyone, I think, is: if your gear is capable of higher sampling rates, TRY IT OUT. If you hear improvement and can afford the performance hit, use it.

Yes, good advice - you always have to find the methods that work for you, rather than just base something on what you read on the internet. If you really feel 96KHz is worth it, by all means I have no problem with that.

all converters exhibit measurable jitter, imperfect LPFs, approximations of infinite sinc functions, etc., and with all due respect to its beauty and elegance, Nyquist-Shannon addresses none of these things. Ultimately, once again, we're left with the uncomfortable position of actually having to listen and decide for ourselves

Totally agreed! Excellent post - thanks James!

Hi,

48Khz - 24bit is the best compromise to get high quality and processing power
If you make music with Samples based instrument 44.1 should be fine too (All most sampes are at 44.1...

But 48 is better for Video Music DVDs!

So I suggest to use 48Khz!

if you don't use Samples, Loops and so on... if your recording session has made only by using Live recording instruments in a very perfect "Capture sounds room" with super High quality Mics and Analog preamplifier connect to a Synphony Apogee system you can start to think to operate with Higer Samples rate!

In any case the actual AD/DA converter (Motu standard quality) just work with Oversampling in the conversion phase... so 44.100 recorded from a standard quality 2009 converter is fully capable to cover all humans range...

There is not a Human with blind test that is capable to ear difference of the same recording at 44.1 and 88.2 or 96... if do a blind test!

Maybe you can find a difference only with sofisticated software that reveal something that is NOT UDIBLE by Humans in any case!

G

I suggest to record everything at 22.05 khz 8 Bit, then play it through a Marshall amp at full blast, record the result at 192 khz 24 Bit, convert it to 88.2 16 Bit khz and rerecord that through an old Behringer desk with very noisy preamps to 44.1 khz 24 Bit. Then convert the result to 96 khz 24 Bit and finally bounce that to an MP3 at 96 kBit/s, carry the MP3 around the studio location at full moon several times (clockwise) and you will be stunned how amazingly ****** it will sound!

I'm sure of that!

in this case you can use also Samples at 44.1 without problem!

🙂

G