88.2khz versus 96khz? (24-bit)

So I'm trying to make my workflow as consistent as possible and in order to minimize problems and distract myself with uncertainty and spending time converting audio, my goal is to set personal standards for two things:

 

1. Sample rate in my DAW for the music I'm making.

 

2. Sample rate for recorded music I'm digitizing from cassettes and vinyl.

3. Any advice on what to use for converting higher sample rates to lower rates.

I have Ableton 8.2 linked to a Tascam US-122L interface using ASIO. For ripping I'm thinking of using Audacity, though I question if it's the best choice, and if so I might stick with Ableton or acquire Reaper for that task. I have a friends copy of Sony Acid but from a superficial first try I don't really like it.

So the Tascam interface can go up to 24-bit, 96khz which and I'm perfectly content at that being the max (no 192khz for me). So my questions are:

1. Is 96khz better than 88.2khz? I'm sold on 24-bit but I can't tell if there's a benefit of 88.2 in conversion going to 44.1 versus 96 to 44.1 (for CD's). Is it better to use 48 and 96 because it's also used for video sound and digital hardware sample rates?

So basically, is there a difference between 24/88.2 and 24/96?

2. I see a lot of vinyl rips done at 24bit/96khz. I know with few exceptions cassettes are inherently lower quality (metal type IV is probably better than cheaply-mastered vinyl). Is 24bit/96khz overkill for cassettes and vinyl? I'm pretty sure 16/44.1 is just fine for archiving my rare and unique cassettes but not so sure about using 16/44.1 for vinyl. After all, digitizing is still, well digitizing…right? Will 24/96 even make a difference in playback of pre-recorded music?

3. Is Ableton adequate/ideal for dithering and converting? I'm committed to recording and mixing with it because of familiarity and my workflow thus far. But should I export the raw mix and use something else for better conversions to lower rates?

Also, this might be the dumbest question, but once I set Ableton to my Tascam for audio input/output and set ableton to 24bit/96khz does everything I record with my Tascam UL-122L have to be in 24/96 no matter what? Or could I set the Tascam (using it's driver preference screen) to 44/16 for my recordings of analog sources but keep my Ableton project set to 96khz? To be honest, this is where I'm most unsure. I don't want to keep changing my interface to different rates when I record different things if it can affect the sound quality within the same Ableton project.

Fuck, this is long and way too OCD sounding. Sorry guys. Feel free to turn this into sample rate advice thread too, after all this one is a bit old: http://forum.watmm.com/topic/8241-sample-rate-bit-depth/

Edited February 8, 2013 by joshuatx

1. use 88.2 if you really must (easier to convert to 44.1), you can't hear the difference to 96 nor 44.1 btw. the only benefit in compaction to 44.1 is that 88.1 audio will be better processed by processing plug-ins but in 99,9% of cases you will not hear the difference. there's lots of plug-ins with oversampling modes now btw so no need for general high sampling rate in your daw. watch only for the best ones for this.

 

2. use 16/44.1 the rest is overkill if no additional processing needed

 

3. use Voxengo r8brain PRO noooooot Ableton for this

 

here's Voxengo r8brain free too

Edited February 9, 2013 by xox

here is a sample rate conversion comparison website for different audio softwares http://src.infinitewave.ca/

 

i'm not sure what it all means

 

you could work in your daw at 44.1 but render at 88.2 to see if you like the way DSP sounds better. I'm still cozy with 16 bit 44.1 from start to finish.

http://people.xiph.org/~xiphmont/demo/neil-young.html

best quote to take from that is the outro

  Quote

 

Why push back against 24/192? Because it's a solution to a problem that doesn't exist, a business model based on willful ignorance and scamming people. The more that pseudoscience goes unchecked in the world at large, the harder it is for truth to overcome truthiness... even if this is a small and relatively insignificant example.

i attended an audio engineering school and not one of my teachers ever admitted to this, they all bought into the industry push for 'high fidelity'.

^
Even while as I'm still learning about all of this it's been clear even to me for awhile that 192kps is overkill even as a recording sample rate. It's being touted by commercial formats like DVD-Audio and Super Audio CD after all.

I love Neil Young to death but I feel like he someone told him analog master tapes have a digital equal in 24bit/192khz recording and he's stubbornly stuck with as a fact. It'd be cooler if he just got behind the benefits .flac or .wav over .mp3/.acc and promoted that.

  On 2/9/2013 at 12:21 AM, xox said:

1. use 88.2 if you really must (easier to convert to 44.1), you can't hear the difference to 96 nor 44.1 btw. the only benefit in compaction to 44.1 is that 88.1 audio will be better processed by processing plug-ins but in 99,9% of cases you will not hear the difference. there's lots of plug-ins with oversampling modes now btw so no need for general high sampling rate in your daw. watch only for the best ones for this.

2. use 16/44.1 the rest is overkill if no additional processing needed

3. use Voxengo r8brain PRO noooooot Ableton for this

here's Voxengo r8brain free too

That program looks perfect - exactly the recommendation I was looking for.

  On 2/9/2013 at 12:30 AM, RadarJammer said:

here is a sample rate conversion comparison website for different audio softwares http://src.infinitewave.ca/

i'm not sure what it all means

Nice, I saw that earlier on another site. It's FAQ and Help menus explain how to compare graphs quite well, though actually comparing all of them seems a bit daunting.

Well between these responses and other things I've read today here's where I'm at:

1. I know there's no audible difference, but I still want to err on the side of caution and go with 24/96, based on the idea that it's better for running vst and plugins. Before this thread's answers I had no idea that oversampling hasd improved so much though - fyi I'm using mostly free vsts recommended here (TAL for example) and Ableton's plugins. Still could use more advice/feedback on this, mostly out of curiosity - I appreciate the feedback so far. Also, if any Ableton users have any advice, especially anything regarding recording and settings, that would be swell.

2. Answered! - Gonna stick with 16bit/44.1khz for ripping cassettes and vinyl. Will match my archive of .wav rips of my CDs as well. There isn't really anything to capture with a higher sample rate, after all CDs were successful originally partly because they eliminated tape hiss and could capture higher frequencies better.

3. Answered! - Will not use Ableton for sample conversions down to CD and mp3 but instead export it raw and go from there. Edited February 9, 2013 by joshuatx

DVD-audio is at that sampling rate but it's going to sound the same if you do a 5.1 surround sound mix at 44.1khz 16-bit.

those commercial audio formats were designed, at least in my mind for people 'afraid' of digital in general to attract audiophile types (like my father in law who has a subscription to stereophile magazine). Most of the releases are of classical music and classic rock.
My father in law is also afraid of even touching mp3 or flac, he assumes that they must inherently sound worse.

a digital audio format will never sound like 2-inch tape (which is what im assuming Niel young is talking about). The only real reason to record to tape is because of the harmonic roll off effect it causes, especially when pushed into the red (tape distortion sounds far more pleasing to the ears than digital clipping). There is no possibility of getting this same effect even if you went up to 192khz, the only time digital is going to come close to sounding like that is when someone creates a perfect analog tape modeling algorithm. Waves and various other companies have tried, and they sound pretty close.

Edited February 9, 2013 by Awepittance

 

  Quote

Also, there are (and always will be) reasons to use more than 16 bits in recording and production.

"In theory, a Nyquist frequency just larger than the signal bandwidth is sufficient to allow perfect reconstruction of the signal from the samples: see Sampling theorem: Critical frequency. However, this reconstruction requires an ideal filter that passes some frequencies unchanged while suppressing all others completely (commonly called a brick-wall filter). In practice, perfect reconstruction is unattainable. Some amount of aliasing is unavoidable.

Signal frequencies higher than the Nyquist frequency will encounter a "folding" about the Nyquist frequency, back into lower frequencies. For example, if the sample rate is 20 kHz, the Nyquist frequency is 10 kHz, and an 11 kHz signal will fold, or alias, to 9 kHz: . However, a 9 kHz signal can also fold up to 11 kHz in that case if the reconstruction filter is not adequate. Both types of aliasing can be important.

When attainable filters are used, some degree of oversampling is necessary to accommodate the practical constraints on anti-aliasing filters: instead of a brickwall, one has flat response in the passband up to a point called the cutoff frequency or corner frequency, (pass all frequencies below there unchanged), then gradual rolloff in a transition band, finally suppressing signals above a certain point completely or almost completely in the stopband. Thus, frequencies close to the Nyquist frequency may be distorted in the sampling and reconstruction process, so the bandwidth should be kept below the Nyquist frequency by some margin (frequency headroom) that depends on the actual filters used.

For example, audio CDs have a sampling frequency of 44100 Hz. The Nyquist frequency is therefore 22050 Hz, which is an upper bound on the highest frequency the data can unambiguously represent. If the chosen anti-aliasing filter (a low-pass filter in this case) has a transition band of 2000 Hz, then the cut-off frequency should be ≤ 20050 Hz to yield a signal with negligible power at frequencies ≥ 22050 Hz and complete pass of frequencies ≤ 20 kHz (within the human hearing range)."

There is the argument of the higher frequencies you can't hear affect the ones underneath but again you need some serious equipment to challenge this theory and sad reality in todays world is most listeners are happy to hear things in mp3 format on shitty white headphones.

16bit offers -92.7db dynamic range, 20bit -103.6db and 24bit -144db but the best D/A converters are around 105db at best and higher than 16bit is only really advantageous in a recording studio environment where you have to avoid the noise floor from hardware which isn't too much of a concern in a DAW.

Edited February 10, 2013 by soundwave

(null)

Bob Katz says 48kHz/24bit is probably good enough.

 

Dan Lavry says 96kHz/24bit is well good enough.

 

They're both pros.

 

-S

I think you're better off working under the standards of your end format. Also, it doesn't really matter.

My mastering engineer, Jorg Burger (of Burger/Ink fame) has told me explicitly that 24 or 16bit at 44.1Khz is totally fine for digital and vinyl mastering. He never wants the artists to send 96 khz files. that guy definitely knows his shit.

i think the best free sample rate conversion you can get is SoX. you can see that the graphs for it 'look better' than those of r8brain at the site that radarjammer linked to. how much the difference may be noticeable is up in the air, but i switched from r8brain to SoX because why not? even if the difference is super subtle, why not make the switch? at least, that was my logic. they are both free. the only thing abou SoX is that it is a bit less wieldy to use. it's command line. UNLESS you have foobar, then you can get a foobar plugin to use with sox and then you can just drag and drop stuff into a foobar playlist, and convert it all at once. easy. of course you have to set up that plugin (only the first time. you can create presets, so you would make one for 44.1kHz, and/or any other rates you might ever want to convert to) and that takes a few minutes to figure out.

Edited February 12, 2013 by MisterE

and as far as the sample rate not mattering, i think it does, at least if you are talking about the sample rate of a project you are working on, where many processes are going to be done to each element in a mix or whatever. sure many plug-ins do oversampling now, but not all. so really the main issue there is about aliasing. even if a plug does oversampling, does the next one in the chain do it? if you work at a higher sample rate when applying long plug-in/effect chains, then that's like you are forcing an oversample. then after you are done you can convert down. but even if a plug, or lets say EVERY plug-in in an 8 effect chain uses oversampling and you are working at a lower rate, that would mean you are up and down sampling that audio for every one of those 8 plugs. thats 16 sample rate conversions.

 

is aliasing that big of a deal? some people claim you can't hear it. i think it depends on how much plug-in processing you do and even what types of stuff you like to use. it will add up. say you run a track through a boat-load of processes. eventually that aliasing might start masking the actual sound and making things seem muddy. people can argue that this is bogus or whatever all they want, but you can see aliasing on graphs, and if it gets loud enough, a masking effect WILL occur. so now lets say you run 44.1khz audio through 8 plugs that DON'T use oversampling, and some of them are things like distortion fx. you're going to have some aliasing that may make your sound a little more muddy, or less defined, or it may have strange resonances that you didn't want.

 

if you are just talking about converting a single file such as a ripped vinyl or something, that issue isn't really going to apply much at all, unless you are going to be doing lots of processing to it. and even in the worst case scenario with aliasing, it comes down to how much you care. you don't have to care. maybe part of your aesthetic is to just do shit and have it come out rough and raw, and that's fine. but to say that these things don't exist at all and/or will not have any effect on anything, under any circumstances, is just plain wrong.

i think it just comes down to one's personal preference of perfectionism. If you want to achieve the highest (technical, not audible) quality possible then by all means set your session at 192. If you have a lot of hard drive space to play around with then keep in mind that 96khz files will be roughly double the size, data wise as a 44.1khz file. If you don't use SSD or ram-disk the loading time when opening and saving these files will also be twice as long. For me, being an impatient music maker, having a minute inaudible improvement over the waveform is not worth the extra hard drive space, the conversion time to turn it into a listening copy and the extra time it takes to save
I've heard this theory that plugins sound better when you operate them at a higher sampling rate, but the truth is a lot of these plugins lie about what sampling rate they are actually capable of (inside the coding). I've actually had to sign NDAs with companies that implicitly say that if i publicly reveal that the plugin they made actually processes audio at 22khz (while it claims on the box it can process at 96khz) i could be sued. I can't prove it, but i'd be willing to put money down that a lot of these claims of being able to process audio differently above 44.1khz is mostly smoke and mirrors.

That link someone posted above also emphasizes (correctly) that unless you have equipment at the end of your chain that can properly broadcast these sampling rates, there is actually a chance of *worse* quality because of various artifacting that occurs than using traditional 44.1khz.
This can't be ignored. If you're going to go above 44 or 48, you absolutely need to have equipment that can handle it.

in response to MisterE, i don't think anyone here is saying that they don't have an effect on anything, because clearly they do when analyzed on a graph or when you dig into the data. The point is are they audible? Can you actually tell the difference between and A vs B test of 10 plugins in a chain processing audio at 44.1hz vs 192khz? I don't think most people can, and if they claim they do it honestly can be attributed to the placebo effect also mentioned in that article.

Edited February 12, 2013 by Awepittance

yeah i mean i dont know for sure if anyone would or even if i would, hear the difference in such a test. it's an interesting idea for a test but i'm sure it's been done at places like gearslutz. but looking at graphs and seeing where the aliasing from harmonic distortion can come up to in level, i would just think that logically, it is going to interfere with the uhh, 'main' part of the signal that you want to hear.

 

but yes i will concede that many people or hell, maybe even most people (myself maybe being one of them) couldn't hear the difference. on the other hand maybe you or i would if actually tested. with my stuff i've been working at 96kHz but it's just kind of because i have a nice system with some good power, and disk space is pretty cheap these days so.. i mean, i would *think* based on graphs i've seen, that aliasing would add up and become noticeable, but i have to admit that i've never tested, at least not comprehensively.

Edited February 12, 2013 by MisterE

If you use virtual synths and software saturation / distortion / non-linear compressor / effect, go for the highest bit depth / sample rate your computer can run, from the very beginning and stick to it until tune is finished.

 

depending on the kind of synthesis and processing, you can hear some significant differences between sample rates. wether you hear it / appreciate the difference is of course up to you.

 

don't forget, SRC and dither happens only once : once the tune is mastered.

 

it's not about the final track, any proper mastering engineer will now how to do the SRC and dithering to the targeted medium.

 

It's about what sounds best to you when you're composing/producing/mixing. use whatever sounds good and don't worry about anything else.

quote system is officially pissing me off today - fuck

 

MisterE:

  Quote

 

i think the best free sample rate conversion you can get is SoX. you can see that the graphs for it 'look better' than those of r8brain at the site that radarjammer linked to.

 

  On 2/12/2013 at 9:18 AM, lin said:

don't forget, SRC and dither happens only once : once the tune is mastered.

 

it's not about the final track, any proper mastering engineer will now how to do the SRC and dithering to the targeted medium.

 

This is good too remember, nonetheless I'll check out SoX as well as r8brain - I've seen both recommended elsewhere. And as I mentioned earlier, I'm digitizing my analog tapes and vinyl in 16bit/44.1khz and more often than not going to be using those or other 16-bit/44.1khz audio files when using samples. I'll convert anything else I import as needed. So questions #2 and #3 are answered.

 

Slowly but surely I'm feeling more and more informed about sample rates. I've been lurking in Ableton's forums, but sorting through misinformation and fact is a lot harder over there (like this clusterfuck thread) than here.

 

It really seems that virtual effects plugin and virtual synth sound quality is the main focus of my sample rate decision. After all, I'm not really recording anything "live" (i.e. no hardware) but instead running midi I sequenced and samples I've arranged within the project. I suppose I can just switch the sample rate from 44.1 and 96khz in the preferences screen on Ableton and render each one to test...right? If that's the case testing should be pretty easy.

 

And lastly, 44.1 versus 96 seems pretty negligible, but is 16 versus 24 bit a more relevant debate regarding plugins and effects than 44.1 vs 96khz?

 

Sorry about my continued questions - I hope it's clear I'm narrowing down my considerations, this thread has been very helpful.

Edited February 12, 2013 by joshuatx

my personal take is that sampling rate probably has more of an effect on plugin processing, where as bit-depth has more of an effect on recording, but the effect won't be noticeable at all with recording of electronic instruments, more with acoustic or field recording.

  On 2/12/2013 at 11:14 PM, Awepittance said:

sampling rate probably has more of an effect on plugin processing, where as bit-depth has more of an effect on recording

 

definitely. and soft synths will almost always benefit from both high bit depth and sample rate.

 

 

joshuatx : go for 24bit when digitalizing your vinyles & cassettes (and when recording anything actually) : no need then to go into the "red" to have a recording with huge dynamic range (144dB at 24bit instead of 96dB at 16bit) and healthy peaks.

 

most A/D converters inputs will behave better if you feed them with moderate levels, close to their optimal level (depending on the soundcard/converters, -24dB / -20dB or -18dB RMS). some engineers even advocate recording at 24bit with peaks at max. -16dB and claim to hear a significant differences, in a very good way. give it a try !

 

 

 

  On 2/12/2013 at 11:14 PM, Awepittance said:

the effect won't be noticeable at all with recording of electronic instruments, more with acoustic or field recording.

 

whenever going digital, headroom is crucial. at 24bit, no need to maximise every bit available nor to stress A/D converters by running them out of their confort zone. any recording will benefit from high bit depth.

 

if electronic instruments aren't as expressive / nuanced as acoustic ones / field recordings, blame it on the human patching the synth/playing it, not on the synth ;)

  On 2/13/2013 at 9:54 AM, lin said:

 

  On 2/12/2013 at 11:14 PM, Awepittance said:

sampling rate probably has more of an effect on plugin processing, where as bit-depth has more of an effect on recording

 

definitely. and soft synths will almost always benefit from both high bit depth and sample rate.

 

 

joshuatx : go for 24bit when digitalizing your vinyles & cassettes (and when recording anything actually) : no need then to go into the "red" to have a recording with huge dynamic range (144dB at 24bit instead of 96dB at 16bit) and healthy peaks.

 

most A/D converters inputs will behave better if you feed them with moderate levels, close to their optimal level (depending on the soundcard/converters, -24dB / -20dB or -18dB RMS). some engineers even advocate recording at 24bit with peaks at max. -16dB and claim to hear a significant differences, in a very good way. give it a try !

 

 

 

  On 2/12/2013 at 11:14 PM, Awepittance said:

>>the effect won't be noticeable at all with recording of electronic instruments, more with acoustic or field recording.

 

whenever going digital, headroom is crucial. at 24bit, no need to maximise every bit available nor to stress A/D converters by running them out of their confort zone. any recording will benefit from high bit depth.

 

if electronic instruments aren't as expressive / nuanced as acoustic ones / field recordings, blame it on the human patching the synth/playing it, not on the synth ;)

 

 

This is actually quite helpful guys - I didn't even think about that aspect of bit-depth with recording. For digitization of my music collection I'll stick with 16/44.1 but for my music recording in 24-bit sounds better for flexibility sake. And I actually will probably switch from 24/96 and 16/44.1 depending on what I'm actually planning to utilize for an individual track.

 

This has been incredibly useful information everyone, thanks!

Edited February 13, 2013 by joshuatx

A/D conversion at a higher sample rate will level up your SNqR which is a good thing... the quantization errors due to the converters will be extended trough a wider bandwidth so, when it lowpasses your signal at 20 kHz all that quantization noise will be filtered...

 

maybe some native english speaker will explain this better...

i think it's called oversampling...