48K vs 44.1 - That big a deal?

[Loudnoize Ent.]

Tell me fella's, are there any advantages of recording at 48k and ultimately ending up at 44.1? The difference between 48 & 44.1 is just short of 4000 samples per second and if I'm correct in saying this, a higher sampling rate means an extended frequency response, right... Well, is there actually a aural difference and also, when converting to 44.1 from 48, is there degradation?

Some insight would certainly be appreciated. Thanks guys.

[Mount Royal]

Don't start in 48 if your end product will be 44.1.

Discussed at length, although years ago as there are now newer sample rates to debate - Concensus clearly states that sample rate conversion from 48 to 44.1 is gravely unhelpful, and that resulting degradation is wildly out of proportion to any benefit that the higher initial resolution that 48 offered. Many advise, in fact that you not sample rate convert from 48 to 44, but that you instead leave your 48 environment analog, and enter your 44.1 environment with a brand new AD conversion.

As sample rate have risen, though, the benefits of the higher resolution at the capture stage are greater. As you might guess, the concensus seems a little less clear presently about how to manage the trade-off between high capture sample rate and the baggage that high SR bring - especially when your final output is CD audio (or worse). If I'm right, there are at least theoretical benefits of working in integral multiples of 44.1 (2x44.1, 4x44.1) at the capture stage, yileding more uninterpolated results at the SRC stage. Others who know much more than I do about this will chime in I'm sure. All of our work is still done at 44.1 with CD audio as the end result.

Best,

John Caldwell

[OgnKog]

I've recently moved to HD2 and I'm also wondering about this... will working at 88.2 be better, even if I have to end at 44.1? Or, as stated, should I start at 44.1 and forget the rest.

Thanks,

Al.

[mixer]

if you want to have material ready for the newer formats..dvd or hdcd i would record at those sample rated..if not 44.1 or 48.

[lancejorton]

first, i completely agree with John. you'll lose more in the 48 to 44.1 conversion than you gain by recording at 48. i would definitely go analog and then A/D back to 44.1 if i got something at 48 (unless i had a really nice converter like a Weiss).

on the 88.2 issue, i think 44.1 ideally should be just fine. but because of error rates, jitter, etc. (real world issues), i think 88.2 sounds more open than 44.1. i probably see more of this since i use an external summing buss, but it really opens up the sound of the instruments... particularly dynamic instruments like drums. another reason to use 88.2 is that the internal summing and routing sounds somewhat better in PT. but i would just do some test runs and decide for yourself.

--
lance

[c-tone]

Well, you are going to get a lot of varied responses on this, and it has been discussed to death on this forum, so please do a search. Many people can't hear a difference between different sample rates and many can. Some people say it is impossible that there is a difference, but I have been able to tell the difference between 44.1 and 48, and I can hear a huge difference using 88.1 or 96K over the lower rates, at least with ProTools systems.

Pretty much everyone agrees that you should avoid any sample rate conversions, but if you have to do a SRC, the conversions that are simple factors of 2 are the easiest and least detrimental to your sound; 96K to 48K for instance. The point is, figure out first what your main end product will be, and then you can make a better decision. For example, if you are recording a music album for CD, your primary end product is at 44.1. If that's the case, then here is what I recommend: if you know you will be mixing within ProTools with no or very little additional analog conversion (outboard gear, consoles, etc.), then record at 44.1. You will be able to have a lot more tracks and plug-ins that way. If you plan on mixing on an analog desk or with a lot of analog outboard gear, then use higher rates like 88.1 or 96K. I personally like to use 88.1 since there is very little difference sound wise between 88.1 and 96K, and when I mix back into two tracks on the ProTools, the Sample rate conversion to 44.1 is a lot simpler, and supposedly will be of better quality. Some techs at Digi say that all sample rate conversions on their gear will be the same, but I like to stick with the simple calculations if at all possible.

[Kuba Pietrzak]

From my point of view the choice between 44 and 48 seems to be very easy...
If you do anything for Cd, radio and so on do it in 44.1 (or 88.2 or...)
If you do anything for Tv, Cinema do it in 48 (or 96 or..)
These are stardarts...
The sonic caomparison between different sample rates was discussed hundret times on several forums (also on this)

Regards
Kuba

[PTUser NYC]

I agree with everyone - record at 44.1 kHz if the final product will be delivered at that rate.

But here's an interesteing question:

Assuming that the material is going to be mastered, and further that it will include at least one trip into the analog domain. Assume that some sort of analog device, whether it be a an EQ or compressor is going to enter into the chain.

At that point, I wonder if the original sampling rate has much bearing on the sound at all. Of course, we won't hear any source information at frequencies higher than half the sampling rate, but as far as numerical consonance or dissonance between source rate and re-recoding rate are concerned, I wonder if they have any bearing?

Bandwidth aside, how can the digital converter "downstream" tell anything about the converter that preceeded it?

If, for example, I make a digital recording of material that has no energy above 15 KHz, and I make that recording at a sampling rate of 48 kHz, is the resulting file audibly distinguishable from one made at 44.1 kHz? I'm going to say no.

HOWEVER, I do believe that I could distinguish the original 48 kHz file from one that had been SRC'ed down to 44.1 - I'm wondering why.

Is it possible that the math of SRC is imperfect? Is there a substantive audible difference between doing the mathematical file operation, and doing an analog playback of the 48 kHz file into an A/D converter set at 44.1 kHz?

I realize that no converters are perfect, so distortion and noise will be added in both the playback and resampling operations, but I'm going after something different here. Theoretically, if the converters were perfect, how would the operation be different than the mathematical DSP version?

If they would be different, I'm guessing it would have something to do with the reconstruction filter, and the slopes of the curve between sampling points. I wonder what my analog devices do to the audio that runs through them. Tranformers, Tubes, Filters, Amplifiers all change the waveform of material that passes through them. I wonder if any of them overrule any "sonic fingerprints" that give away the playback converter's playback rate.

In a nutshell:

IF a theoretically perfect analog transfer is different from the mathematical transform, then the difference must be in some attribute of the waveform. Do analog devices generally change these attributes significantly enough to create a "new analog signal" which has no limitations (other than bandwidth) as a result of having been once recorded digitally? So that the new recording converter's sampling rate would be audibly irrelevant, other than bandwidth limitations?

If you record a signal which has no energy above 15 kHz, at both 48 kHz and 44.1 kHz, and run each file through a Pultec with moderate settings, recording the result of each at both 44.1 kHz, and 48 kHz, would the resulting 4 files be audibly distinguishable?

If not, then the solution to SRC is somewhere in there.

Someone please enlighten me, and help me understand this better? Nika? Anyone?

[Loudnoize Ent.]

Thanks guys... these are all very, very interesting points. So what most of you are saying is to work at the samples I intend my end result to be in. Some of the pros I know do record at 44.1, as they have mentioned that 4000 samples from 48 when down-converted to 44.1 causes sound degradation. Can we actually hear this? Personally, I have a hard time hearing the difference. Also do mastering houses prefer 48 as opposed to 44.1?

[Steve MacMillan]

To make matters more complicated...

CD are at 44.1k, but many DVD Audio disks are at 96k. So if you are doing a Hifi recording that might end up on DVD Audio or some future format, 44.1 and 88.2 will require non even sample rate conversion where you least want it. Most people working for film record at 48k.

Really I don't worry about it too much. If need be I'm not at all worried about an analog conversion at the mastering level with the best gear. I do think I can hear an improvement at 48k over 44.1k. If you're recording your band at home and will staying digital right thru to the CDs you sell at the door, record at 44.1 or 88.2k.

STeve