R/E/P Community

R/E/P => R/E/P Archives => Bruno Putzeys (Designer) - Dave Hecht (Master Tech) => Topic started by: Dougtune on May 20, 2004, 08:54:25 AM

Title: 44.1k or 48k
Post by: Dougtune on May 20, 2004, 08:54:25 AM
My digital mixer and separate hard drive recorder can run at 44.1Khz or 48Khz, among others.  Is there any consequence at all, for sound quality or future compatibility, in using one or the other setting?
Title: Re: 44.1k or 48k
Post by: josh on May 20, 2004, 09:27:26 AM
Yeah.  44.1kHz will not require any sample rate conversion when you mix the tracks for a CD.  48kHz will.

Title: Re: 44.1k or 48k
Post by: John Klett on May 20, 2004, 10:29:07 AM
yeah - what josh said!

well this is one of those questions that really has no solid blanket answer... for your particular application 44.1KHz probably does make sense... but in other applications other things can come in to play so here are a couple considerations.

If you run at 48KHz the anti-aliasing filters move a little higher and the assumption is that the further away from (higher than) the audio band the filters are the better.  There are a bazillion explanations for why higher sample rates sound better.  Many of these are esoteric...  hard to really wrap your head around unless you can sit down and work through the various mathematical and psycho-acoustical  arguments...  and I expect most of these are quite valid.  A much less esoteric explanation that is easy to understand is that, in practice, anti-aliasing filters in mass market gear (and most of what people use is mass market) are not the best that they can be.  They should be completely out of the way below their cutoff frequency but in practice they do impose artifacts.  When you increase the sampling rates those filters move up in frequency as well...  the artifacts move a little bit more out of the audio band and maybe that is something we hear as an improvement.  If this is true the qualitative improvement moving from 44.1 to 48KHz should be more audible in cheaper and lower quality equipment where the filters would tend to be crappier.  This, in fact, has been my experience.  In very high end equipment the difference is miniscule.  I got to live for a time with a Pacific Microsonics HDCD II AD/DA converter that I used for some mastering work and I really couldn't tell the difference between 44.1 and 48KHz.    ...so that is something...  

When you make the move to 2x sampling rates the differences are more dramatic and noticeable...  going from 48KHz to 96KHz makes an audible difference even in the high end gear.  This makes me think there is something to the claim that we somehow are sensitive to harmonic content that is supposed to be well above our audible band...  but this is a contentious topic

As far as compatibility...  CD's release at 44.1KHz and I know quite a few guys who do everything at 24bit/44.1KHz and simply dither down to 16bits for the CD in mastering.  They avoid sample rate conversion (SRC) and...  again...  here is another thing (process) that can introduce artifacts.  SRC can be really really good or downright cheesy.  Where you fall in this range is often linked to how much you spend.  You can avoid the whole SRC 'thing' by starting and staying at 44.1Khz from top to bottom.  On the higher end...  SACD releases are done at a bit rate that is an integer multiple of 44.1 and when we do work the is bound for SACD we encode into Pyramix system (this would be a multitrack that we are remixing for surround) at 88.2KHz or 176.4KHz.  DVD-A releases can be 1x, 2x or 4x either 44.1KHz or 48KHz so one would think that 44.1 based sample rates would be "better".

Oddly enough, while it is intuitively obvious that SRC up or down in even integer multiples (44.1 to 88.2, 176.4 to 44.1 etc.) would be easier to do than 48KHz to 44.1KHz or 44.1KHz to 96KHz and so on...  I have heard enough argument on both sides that I guess it may not be so obvious.

So...  I work differently depending on the project.  So - back to what josh recomends....  he's right.  At home I have an ever expanding analog setup with a ProTools LE and a 001.  I run at 44.1 so I can mix in the box or loop out and back into a project file and make a CD without having to deal with the SRC issue.  It also saves me a little disc space.  All stuff that I do at home is pretty much going to end up on a CD and get turned into an MP3 and sent out to the world that way if it leaves my studio at all.

If I am doing a project at Sound Bunker (it's a tiny studio in CT where we do some re-purpose and authoring for surround DVD-A and DVD-V) I look at the release format.  Almost all the DVD-A work we've done has been for surround at 96KHz so that's what I encoded in when coming from analog multitracks.  If I can skip a process like SRC I will.

a couple quick things...  if I have any TDM based ProTools based system with enough i/o to mix through an analog desk I'd run at the highest sampling rate practical (probably 48KHz even with an HD system for file exchange reasons).  I'd mix back to 1/2" or a MasterLink with a decent converter on the input at 88.2 or 96KHz because I'd be taking it to mastering with high-end processing and SRC tools to do what needs to be done...   OR...  I am in the box and someone hands me a project already started at 48KHz - well - I have to stay there - and that is the most common sampling rate I come across in PT session files.  The SRC has to be handled post mix in mastering...

The answer to your question is really gotten to by asking and answering some other questions so...  I could blab on some more but I think what I have typed so far may give you some small insight.  A lot of it comes down to trying different things over time and just getting a sense of what works and sounds better to you going through the whole process.

Title: Re: 44.1k or 48k
Post by: Dougtune on May 20, 2004, 04:23:19 PM
Thanks for the info, John.  I'll put 'em at 44.1K and forget about it.  
Title: Re: 44.1k or 48k
Post by: OatBran on May 22, 2004, 02:19:55 AM
Something else to consider....


I took a 3 week class with Mr Rupert Neve (yes, the man and the myth).  In this class, he discussed the mathmatical concepts behind sample rates and the psychoacoustics involved with them.

While he is an adament believer in our ability to discern frequencies above 20k, he also believes in the benefits of higher sample rates do to their increased resolution in the time domain... not just the frequency domain.

With increased sample rates you are getting more samples of the waveform per second.  He believes that this increase in time-based information makes the sound much smoother and realistic to our ears.

Something else to chew on.  Having said that, I still use 44.1 regularly myself.  However, when bumping up the sampling rates, I prefer to go 88.2 rather then 96.  I find the transition to the CD sampling rate of 44.1 MUCH smoother that way due to its higher accuracy in the time domain during its conversion.  Then again, if I listened to Mr Neve, I would never use anything but what he considers to be the bare minimum sample rate for digital...

192k.

Ouch.
Title: Re: 44.1k or 48k
Post by: ssltech on May 26, 2004, 10:36:09 AM
Quick perspective check:

If you play a 44.1kHz recording back at 88.2kHZ, it plays back an octave higher. This is a very simple way of showing that the difference in recorded range amounts to one octave when you double the sampling rate.

If you play back a 44.1kHz recording at 48kHz, it plays back at fractionally more than a half-step higher... (1 seimtone is just under a 6% speed difference...)

3.9kHz sounds like a big number until you realise that it's only buying you a semitone at the top end... and since many manufacturers of earlier digital gear didn;t want to build two sets of reconstruction/anti aliasing filters, with close-tolerance components and all the attendant expense, some even left the same set of (lower cutoff frequency) low-pass filters in place, and spent the extra money making them a little bit better, instead of building two sets of cheaper, worse filters... The net result in these cases was a bit-rate penalty with no benefit save for compatibility...

Nobody's ever convinced me that they can tell the difference between 48k and 44.1 blindfold, though the higher steps between sample rates are rather more rewarding.

Round here, professional projects are done at 48k and multiples. Makes the video guys sweat a lt less when we do it that way. Small-time CD release stuff is done at 44.1k, since it means that we get about 7% more storage space on the drives and no SRCs between the mix-in-the-box and the final CD.

If it's going through an analog mastering stage out if house, I ask the mastering guy, but he usually says he'll take either...

There's no single 'right or wrong' answer, but there's often one which works really well for a given project. -Just think about what it has to synchronise with, and that usually makes the choice for you! Wink

Keith