sample rates - arrrr theres to many of them!

queglay · Post by **queglay** » Tue Dec 05, 2006 10:52 am

this is unrelated to my other topic so i thought id post another thread.

I also have another question. what are all your sample rates in your workflow? i've heard its bad to change them, but since all my stuff in the past has been 44.1k (never been able to go higher) with the new step to go up to 96, 192k, or some ridiculously high number, I'm not to sure what i should bank with. is it bad practice since all my old samples are between 44.1 and 48 to go to another higher sample rate or is it ok as long as they are divisible into one another?

what happens if your project has been done at 44.1k, and you decide to export a loop and 48k or 96k?

i think the sample rate deal is the most complex to figure out in order to retain quality.

Cheers again.

dodgyedgy · Post by **dodgyedgy** » Tue Dec 05, 2006 11:39 am

If the original sample rate was 44.1 then changing it into 96 is NOT going to get you a better quality sound.

And BTW... if you think sample rates are complex then try getting your head around video formats... now thats brain rattling.

fsk · Post by **fsk** » Tue Dec 05, 2006 12:00 pm

44.1 for CD's mp3s, avi mpeg etc.

48.0 for Digital Media like DVD's etc

96 + DVD and HD-DVD stuff.

no need to go above 44.1 if ur just doing music really.

queglay · Post by **queglay** » Tue Dec 05, 2006 1:01 pm

i've been doing animation for about 7 years, and i still cant remember all those new hd formats.

I know that upping the sample rate is not going to give me better sound for samples that already exist. but synth's and new recordings will be wont they? will my old 44.1k and 48k sounds actually sound worse?

arent there new audio formats that are stating to catch on with higher sample rates?

corygilbert · Post by **corygilbert** » Tue Dec 05, 2006 1:27 pm

recordings will improve in quality somewhat as long as your ad converter on your soundcard is converting at the sample rate you are recording at.
ie, if your soundcard is only converting at 44.1 then recording at 96 isn't gonna help. Basically the sample rate is how often the converter grabs a piece of audio for conversion. So 96 or whatever is better but not for existing samples.

sqook · Post by **sqook** » Tue Dec 05, 2006 2:38 pm

queglay wrote:i've been doing animation for about 7 years, and i still cant remember all those new hd formats.

I know that upping the sample rate is not going to give me better sound for samples that already exist. but synth's and new recordings will be wont they? will my old 44.1k and 48k sounds actually sound worse?

arent there new audio formats that are stating to catch on with higher sample rates?

Nope, the sound quality will be 100% the same. Anyone that tells you otherwise is (unknowingly, probably) lying to you.

When you go from a lower sample rate to a higher one, this process is known as upsampling. Basically, extra frames get added to your sample and an interpolation algorithm places them strategically in the wave data. However, the exact position of these extra samples is mathematically "guessed", instead of being actually determined by higher frequency waves that would have normally been part of the higher-sampled recording.

The rule of thumb (aka the nyquist-shannon theorem) is that the sampling rate will accurately capture all frequencies up until 1/2 of the sample rate. Since the human threshhold of hearing is about 20khz (though much lower for most club DJ's, on average

), there's no reason to use sample rates above 48khz.

The exception is if you are recording stuff at 96khz or whatever, and then you want to do filtering or other operations at that frequency and then downsample to 44.1khz.

queglay · Post by **queglay** » Wed Dec 06, 2006 12:37 pm

hmm interesting. thanks for the information.
1. i would have thought though, that if i jump from 44.1 to 48, that it might sound worse because the interpolation is not by a whole multiple.

2. obviously going from 48k back down to 44.1 would loose quality, but would it be worse than recording at 44.1 in the first place because of the interpolation problem?

what about the argument between 16bit, 24 bit, and 32 bit sound? hows that work then?

Thanks again,

noisetonepause · Post by **noisetonepause** » Wed Dec 06, 2006 1:14 pm

sqook wrote:Nope, the sound quality will be 100% the same. Anyone that tells you otherwise is (unknowingly, probably) lying to you.

Not 100 percent! Since:

When you go from a lower sample rate to a higher one, this process is known as upsampling. Basically, extra frames get added to your sample and an interpolation algorithm places them strategically in the wave data.

This obviously changes the waveform. Non in any way my ravaged cloth ears will be able to detect, mind, but the waveform will be changed when you upsample... That's why you should record at a sample frequency that's related to the one you're targetting - ie. if you target 44.1 KHz CDs, it's (supposedly) better to record at 88.2 than at 96... Supposedly. This was probably a bigger issue ten-fifteen years ago when digital audio was still young.

there's no reason to use sample rates above 48khz.

Well... yeah. Anything that goes through a DAC is run through a smoothing filter when cuts off anything above sample rate/2 (as you know!), and there is anecdotal evidence from certain audiophiles that it's better to move this cutoff frequency into the inaudible range.

Personally, I record at 44.1 KHz cos I can't be fucked to change the preference.

-Paws

David · Post by **David** » Wed Dec 06, 2006 1:24 pm

queglay wrote:
what about the argument between 16bit, 24 bit, and 32 bit sound? hows that work then?

Thanks again,

the bit depth gives more precision in the quantzation of the amplitude value. So 16 bit gives 2^16 possible quantization levels between -1 and 1, and 2^32 gives, well, that many quantization levels. I personally feel that there is a more noticed improvement in sound quality perception when increasing the bit depth, as opposed to increasing the sample rate (above 44.1)

i think thats the case anyway...

queglay · Post by **queglay** » Mon Dec 11, 2006 8:01 am

thanks for the info, this is interesting. but if your recording at 24bit, and your sending it out to a cd which is 16 bit, then wouldn't you get some interpolation issues there in the final calculation of the amplitude?

and as for the sample rate issue of recording in 96k and going down to 44.1, im guessing thats not the best thing to do, though it sounds like no one is certain of this.

recording at 88.2 and exporting at 44.1 should be fine though yes?

i think its important to nail any of these workflow issues now. i dont want to come back to some loops in a years time and hear, "no dude, you should have done it like this..."

astroid · Post by **astroid** » Mon Dec 11, 2006 8:13 am

maybe this is a good place to ask: wouldn't recording/working at a higher sample rate make time-stretching and pitch shifting better?

like, theoretically, in a 44.1K recording, everything above 22.05K is folded over, which you can't hear, but if you drop the whole thing one octave, all that stuff from 22.05K comes down well within the audible range. so, big 192K samples could be dropped in a 192K session some four octaves without hearing the effects of the sampling.

and for timestretching, wouldn't all that extra sampling aid the algorithms?

is my logic flawed?

Johnisfaster · Post by **Johnisfaster** » Mon Dec 11, 2006 8:25 am

astroid wrote:wouldn't recording/working at a higher sample rate make time-stretching and pitch shifting better?

yes, because upsampling gives you more "material" to process and will make pitch shifting and time stretching better. robert henke confirmed that one for me some time back when I had a very similar thread as this one.

from my own experience, and I sudjest you do this too. open operator up and give it a good buzzing lead sound, put autofilter after it with a good lowpass lfo sweep and then lives reverb after that with a semi long decay on it. now first render a 44.1 version of it, then render a 96 version of it. you'll easily hear the difference between the 2 files it's not hard to hear. if you pay close attention to the really high frequencies you'll hear that 96 is much smoother.

the problem is that most people think they understand this stuff and haven't actually tried using their ears.

FaX-01 · Post by **FaX-01** » Mon Dec 11, 2006 8:26 am

Some interesting points here.
If I might add ....

1.AD/DA conversion plays a big role. For example a 16bit 44.1 khz top end Apogee AD/DA converter may very well yield better results than a lower end 24bit 96khz converter on the same recording.

2.Bit Depth along with AD/DA converter quality is IMHO far more important than just raw sample rate. For example I'd rather use a good 24bit/48khz quality source over a shitty 24/96 source.

I personally prefer to work at 24/48 myself and don't find a massively huge enough difference to warrant the CPU hit on higher sample rates in general.
That however is just a personal preference.
However I must confess that using higher sample rates in Reason 3 does yield far superior results overall this end.
Generally though 24/48 is a happy compromise this end for all intents and purposes.

astroid · Post by **astroid** » Mon Dec 11, 2006 8:36 am

a good master clock, like a big ben, apperently helps keep these things in line, too.

ikke · Post by **ikke** » Mon Dec 11, 2006 8:53 am

I have a question related to this topic.
Does it make any difference if you record the sound at 96khz and then import that sound into a 48khz Ableton project instead of recording it at 48khz?