48k or 44.1?

Discussion in 'Mastering' started by Sork, Oct 31, 2004.

  1. Sork

    Sork Active Member

    I do the mastering of my home projects in wavelab. how is the converters in it? should i record in 48k and dither down at the end, or should I stick to 44.1?
     
  2. Ben Godin

    Ben Godin Active Member

    hey Sork, the converter in WaveLab is not very good, i'll tell you that for a fact. I think its best to record and mix in 44.1, 24 or 32 bit, simply because if you only have WaveLab, and are doing your own mastering, you'll get much better quality without WaveLab's converter. :cool:

    also, if your recording program allows you to export at 44.1 from a recorded 48, i would go ahead and do that, (like you record at 48 and you bounce at 44.1), otherwise stuck to recordnig at 44.1 with a high bitrate.
     
  3. Sork

    Sork Active Member

    Ok, kind of the answer I expected! :wink:
    What if then, I did my recordings at 48k, and recorded into wavelab with a digital cable (real time), with wavelab set at 44.1? Will this sound just as bad as doing the conversion with wavelab? And will setting wabelab to record at 32bit floating make any difference from recording at 24bits? :oops:
     
  4. Ben Godin

    Ben Godin Active Member

    why would you record into wavelab? another cable? bounce it out of your seq. and then import it into wavelab.
     
  5. Sork

    Sork Active Member

    Because I've heard that the mixdown engine in my DAW (Standalone Roland Vs-2480) isn't very good, so I just skip mixing down on the Roland, and just record my mix to a stereofile in Wavelab! And again, will it sound good if I have a project thats at 48k recorded realtime into wavelab at 44.1?
     
  6. Cucco

    Cucco Distinguished Member

    Sork,

    Your problem seems like a fairly simple one. First though, a little bit of terminology. When changing sample rates, there is no dithering that goes on. That is when you change bit depths. However, what is done is, the program removes, based on a predetermined algorithm, approximately 4000 samples. Very few programs do this well, and for that matter, very few outboard sample rate converters do this well.

    Your best bet may be one of the two following options:
    1. Record in 44.1 to begin with. There will be very little (if any perceptable) difference between 48 and 44.1. Any such difference will be destroyed by the downsampling.

    2. Come out of your DAW in analog, into a mixer with decent summing (such as an inexpensive Soundcraft/A&H) and feed the stereo analog signal into Wavelab. In this method, you can insert some outboard gear inbetween the two DAWs and forget messing with as many plug-ins.

    If you need a couple channels of inexpensive conversion, check out ART's DI/O box - it does a decent job for $150, though it is unbalanced only. I'm guessing by your set-up that you don't want to blow too much more on conversion.

    My $.02

    J...
     
  7. Sork

    Sork Active Member

    Ok then, I'll continue recording at 44.1! Thanks for the helpful answers!
     
  8. Ed Littman

    Ed Littman Guest

    Wrong.... when ever there's a mathmatical calculation the word leng increases & you must use dither. src is not basic math.

    What i would do if i were you, is record at 24/48 burn your mixes to disc & send them off to a mastering house...or use a good converter at 44.1 going into your sound card & capture in wavelab at 32/44.1
    I do this every day & it sounds great.
    Ed
     
  9. Cucco

    Cucco Distinguished Member

    Ed,

    I'm glad you do this all the time, but I'm afraid there technically is no dithering in Sample Rate Conversion. The act of ditheriing, by it's very definition is to add information in small particles to an object so that when it is reduced, the reduction does not appear as nothingness. This simply is not what Sample Rate Conversion is doing. Sample Rate Conversion is the state of taking 48,000 samples per second and transforming it to 44,100 samples per second. The only way to do this is to selectively remove some of the samples (nearly 10%). However, this cannot be done simply by randomly removing samples (despite the fact that this is what some Sample Rate Conversion programs do.) One has to be very selective at removing samples.

    For example, you can remove more samples at lower frequencies than you can at higher frequencies. However, despite this, you simply cannot remove samples in an exponentially repeating pattern; the human ear recognizes this pattern. Therefore, you must do kind of a combination of the both (with a little bit of planning behind it.)

    I'm pretty sure I'm correct on this, since, for my day job when I'm not recording orchestras, I write this kind of software for the military intelligence field - and yes, it's entirely based on algorithmic computations.

    Also, while I whole heartedly agree that the best solution would be to record at 32 bit floating point/44,100 samples per second using a high quality converter, I'm afraid you missed the essence of Sork's message. No offence intended towards Sork, but he's making do with the best equipment he can for the money - note the Roland VS-2480. How is it helpful to people on this forum to suggest a $3000 solution to a $5 problem? Yes, the outboard converter would be the best option - but he's asking for a solution that he may already possess.

    I will agree with you again that outsourcing the mastering will yield him the best results, but why does everyone assume that EVERY project must be sent to a mastering engineer? Maybe the guy wants to learn a little of it himself...maybe he's having a little fun...maybe he doesn't have the funds.

    Please, I urge you, check your facts before you post information, and answer the question at hand, not what you interpret to be the question. You'll find it's a lot more helpful. :)

    Thanks,

    J...
     
  10. Ed Littman

    Ed Littman Guest

    Correct, there is no dithering in src but dithering should be applied afterwords.

    I meant to say when reducing wordlegth you must use dither. when any digital process is applied it expands the word length including src.If you need to reduce the word length from that point you need to dither.

    in your quote it looks like your impling that dithering only happens when changing bit deapths. For those that don't know, you have to apply dithering as a last step, it does not happen automatically during that procces. I also urge you to chose your words more clearly.
    Maybe you meant truncation.

    Thats why all the good src hardware boxes have dithering.

    http://www.weiss.ch/sfc2/sfc2.html
    Altronics Amp

    Lastly i didn't miss his point. you can get a good adc for pretty cheep now a days.
    Ed
     
  11. Michael Fossenkemper

    Michael Fossenkemper Distinguished past mastering moderator Well-Known Member

    All of the top hardware SRC's do have a dither option but that's because of their ability to reduce the bit depth. I don't think that they apply dither as the result of SRC'ing. I could be wrong though. I checked the flow chart on my new apogee rosetta 200 which has a/d/a, src, and bit reduction. In the flow chart, dither is applied before the SRC. Now i'm not sure why the SRC follows the bit reduction, but it does. Logically, I would like to see dither as the very last step in this chain but for some reason it doesn't. Maybe it's the design of this unit only for reasons I don't know yet.
     
  12. Ed Littman

    Ed Littman Guest

    hmm, dither before src.
    i think most of those units have a higher internal sample rates. the weiss proccesses 40bit float/32 fixed, so i don't know first hand but based on that I would guess dither is offered after src.

    I remember on glenns board a conversation about this, & quite a few good engineers missed this point until dave collins & goran finnberg set every body strait . at least the fact that src needs dither afterword.
    Ed
     
  13. Ed Littman

    Ed Littman Guest

    sork,
    if you use wavelab to src, don't use the process src, use the resample 192 plugin in the master section. even pg the designer of the program recomends that plug over the internal one.
    Ed
     
  14. wiz1der

    wiz1der Guest

    ok, but back to the original question....

    What if he comes out of the 2480, into a good summing mixer as mentioned, then out of that into a dat recorder, say a panasonic 3800, to utilize the converters in there?

    Digital out of the 3800 into the other comp.

    I noticed that the panny's were the ones you saw in ALL the studios about 5-10 yrs ago, so I'm guessing the converters are decent?
     
  15. Cucco

    Cucco Distinguished Member

    Ed,

    I think it's important that you understand that dithering and SRC (Sample Rate Conversion) are completely independent of each other. You can put the SRC anywhere in the chain that you would like, as there is no dithering that occurs at this point.

    Also, I said nothing about bit depth reduction or truncation, which is two different processes, so what you are referring to by - "in your quote it looks like your impling that dithering only happens when changing bit deapths" does not seem to make any sense. However, dithering DOES only occur with bit depth reduction.

    When you simply truncate a word from 24 bits to 16 bits, you are, in essence removing not just the extended dynamic range of the recorded work, but you are also changing how the digital hardware deals with the analog voltage. In a 24 bit word, the incoming analog voltage (given in whole integers with an infinite number of decimal places following) are represented a binary number representing that voltage, up to about 4 decimal places. In a 16 bit word, you get the same principle, but a representation of up to only about 2 decimals. Therefore, the transition between sounds of different amplitude is much less smooth - almost represented as a stairstep effect if graphed.

    While you can simply truncate the bit depth (easily done by simply patching a 24 bit signal (source) into a 16 bit medium (destination) with no intermediary hardware to apply dithering) it will produce detrimental sounds easily audible to the average listener. However, by applying a shaped noise, similar to what is going on within the program material (and often derived from it), you fool the human ear into thinking it hears smooth transitions. This is dithering and has nothing whatsoever to do with changing the sampling rate.

    The reason that the two pieces of hardware that you mentioned include dithering is because they will perform both sample rate conversion (important!) and bit depth reduction (NOT TRUNCATION).

    Many will argue which of these items needs to come first - SRC or bit depth reduction (often referred to somewhat incorrectly as word length reduction). By placing the SRC first, you remove much of the audio material required by the dither device to create the artificial noise. However, if you place the SRC second, you may actually be removing samples of the carefully created dithering noise as well as program material. Neither of these is the best solution. That is why several people prefer to mix and record in the same frequency that will ultimately be used in playback.

    Of course, I can point out one other obvious point. When recording at a sample rate that is a multiple of your final destination frequency, the Sample Rate Conversion that takes place is simply a removal of evenly spaced samples. For example, when recording at 88.2Khz and down-sampling to 44.1 Khz, you simply need to remove every other sample. You will still get a very accurate picture of the original wave form, with only minor discrepencies at the highest of frequencies (As indicated in the Nyquist Theorum.)

    I think if Dan (Weiss) were to chime in (I understand he lurks around here sometimes) he would agree that even his sample rate conversion follows these basic rules.

    Thanks,

    Jeremy :)
     
  16. Cucco

    Cucco Distinguished Member

    Wiz1der:

    You're thinking creatively, and that's good. However, I don't think that this is the best solution.

    How about this:
    Come out of the 2480 in analog
    Go into a good sound card via analog (RME or Lynx - the Lynx is big $$$, but worth it. The RME can be had for little $$$ and is a good contender. Hell, you can even run into an Echo Mia)

    -OR-

    Come out of the 2480 in analog into a decent converter
    Decent converter sent back into pc in digital in 44.1/16.
    This is essentially what you said, but I don't understand the need to use the converters on the DAT. They're not THAT good.

    -OR-

    Come out of the 2480 in analog
    Go into a decent outboard effects box (Lexicon MPX550, TC digital compressor) which will send the signal to the PC in digital.

    You just have to be careful not to go crazy with the effects, cuz you can't undo them.

    These are just a couple of options based loosely on your suggestion. Of course, all of these are IMHO.

    Thanks,

    Jeremy :)
     
  17. Ammitsboel

    Ammitsboel Member

    Wrong Ed!

    I would be carefull about sending a mastering engineer 48k when I've not spoken with him about it.

    Best Regards
     
  18. Ed Littman

    Ed Littman Guest

    Thanks ,I understand that src & dithering are different procceses, but that doesn't mean you should not dither after.

    src will expand the word legth(32float etc.).
    when ever you reduce word length you dither(back down to 24 or 16bits)
    your qoute was not clear of that fact & so i'm trying to clarify it.

    also, now a days all sample rate converters up sample to a very high frequency around 15mhz thats an even multiple of all of the other sample rates involved. Then is down sampled to the destination sample rate using even multiples reducing any funky math. So it does not matter if you src from 96 down to 44.1.

    upsampling> reconstruction filter>high sample rate>anti-aliasing filter> down sampling

    Ed
     
  19. Ed Littman

    Ed Littman Guest

    [quote="Ammitsboel
    Wrong Ed!

    I would be carefull about sending a mastering engineer 48k when I've not spoken with him about it.

    Best Regards[/quote]

    What do you mean by that. I did'nt say he should not speek to the mastering engineer first.
    Did I miss your meaning here?
     
  20. Ammitsboel

    Ammitsboel Member

    What do you mean by that. I did'nt say he should not speek to the mastering engineer first.
    Did I miss your meaning here?[/quote]

    It just didn't came out clear in your message, that's all.
    By looking at your message he could think that it was ok to deside for himself rather than deside together with the ME.

    Best Regards
     

Share This Page