What bit and sample rates to use?

Discussion in 'Recording In Progress' started by ADinNYC, Jan 9, 2013.

  1. ADinNYC

    ADinNYC Friend of Leo's Silver Supporter

    Posts:
    2,474
    Joined:
    Jan 2, 2010
    Location:
    New York City
    I may get the RME Babyface interface and it records up to 24/192. I know this has been addressed here before but I had trouble finding the threads. If I remember correctly recording at 24-bit is beneficial but increasing the sample rate more than 48 or so really doesn't add much benefit. Anyone care to chime in?

    Thanks

    Andy
     
  2. 64Strat

    64Strat Friend of Leo's Vendor Member

    Posts:
    4,756
    Joined:
    Oct 6, 2003
    48kHz sample rate and 24 bit word depth is all you will need. Anything else just uses huge amounts of disk space.
     
  3. Geoff738

    Geoff738 Poster Extraordinaire Silver Supporter

    Age:
    56
    Posts:
    7,670
    Joined:
    May 11, 2007
    Location:
    Toronto
    Some plugins supposedly sound better at 88/96.

    Personally I'm running 24 bit at 44.1 most of the time.

    Cheers,
    Geoff
     
  4. Old Cane

    Old Cane Poster Extraordinaire

    Posts:
    5,835
    Joined:
    Sep 7, 2007
    Location:
    Murfreesboro, TN
    Well, more is better but comes at a cost. If you can afford it go for it. If you do '60s stones type stuff you probably won't need it but if you're recording classical with some better preamps and great mics it will. I've always used 24/44.1 and been happy and I hated moving away from tape. I realized it wasn't digital I didn't like, it was what people were doing with it.
     
  5. peteycaster

    peteycaster Tele-Meister

    Posts:
    419
    Joined:
    Jul 26, 2008
    Location:
    Australia
    24/48 here. Just make sure your computer has the oomph for it.
     
  6. kp8

    kp8 Friend of Leo's

    Posts:
    4,981
    Joined:
    Aug 22, 2004
    Location:
    Seoul, Korea
    Don't use 48. There is no befit over 44.1 and your software will have to resample for CD and resampled 48k is worse (often much) than 44.1. Resampling well is hard and computationally expensive. Most commercial software packages have mediocre or terrible resampling algorithms. The tiny bit more sample resolution you get at 48 is overwhelmingly negated by having to resample and the artifacts that result from bad filters and crude resample routines. If your final destination is CD avoid 48k.

    Higher bit depth is a huge help. Much more important than higher sampling rates. 20, 24 or 32 bit is a big help over 16 bit.

    I do a lot in 24bit w/ 44.1k.Hz if the final destination is a red book CD.
     
  7. woodman

    woodman Grand Wazoo @ The Woodshed Gold Supporter

    Age:
    74
    Posts:
    15,491
    Joined:
    Nov 28, 2004
    Location:
    Mint Hill, NC
    I used to go 48/24 all the way. But then I realized that 99% of my recordings went to CD or the internet, so I switched over to 44/24 and couldn't tell any real difference from downsampled 48 (though maybe some people with good ears and experience might). The only exception was a song I sold for a film — 48/24 is the standard for soundtracks.

    But as Kp8 says, higher bit depth is a plus ... I've found recording at 24 gives you a good bit more headroom.
     
  8. 64Strat

    64Strat Friend of Leo's Vendor Member

    Posts:
    4,756
    Joined:
    Oct 6, 2003
    While that may be true to a limited extent, it is avoidable. I've done quite a few projects at 44.1kHz (24 bit) and I always like the ones done at 48kHz (24 bit) better. In my own experimentation with all of my different playback systems, when I record acoustic instruments, I can hear a perceptable improvement in the at 48kHz sample rate version vs the 44.1kHz samples I've done. Even when being converted back to Red Book standard with the software I use. My thinking is that it's due to another incremental reduction of quantization noise at the higher sample rates, which is independent of Nyquist sampling theory concerns. Reproducing complex amplitude modulated waveforms with lower quantization noise artifacts is the reason for using higher sample rates. I would also think that higher sampling rates also reduce anti-aliasing filter artefacts. So, 44.1 doesn't quite get me there sometimes, 48kHz is an improvement for me on some things and if CD is the final destination, I would use 88.2kHz sampling because the down sample conversion is a simple integer. If the final destination is mp3 et al, none of this discussion means anything then. :lol:
     
  9. SacDAve

    SacDAve Poster Extraordinaire Silver Supporter

    Age:
    68
    Posts:
    7,067
    Joined:
    Dec 23, 2009
    Location:
    Rocklin Ca.
    I just watched a free video on Lynda for setting up Pro Tools. they suggested for Audio 16 bit for Video 24 bit. It also explained the higher the Sample rate the better the quality but more hard drive is needed. My PT is set at 24 & 41.5 it seems to work fine but I might try a higher sample rate.
     
  10. dwlb

    dwlb Tele-Holic

    Posts:
    660
    Joined:
    Aug 18, 2010
    Location:
    Chicago
    I've been running 24 bit/48kHz at the work studio for a few years and it works well. Only time I go to higher sample rates like 96kHz is when I'm doing sound-design stuff and will need to do extreme pitch-shifting. Most modern computers won't have trouble with the math converting from 48k to 44.1k for your CDs. Of course, it's 2012, why are you still making CDs? :cool:
     
  11. octatonic

    octatonic Poster Extraordinaire

    Posts:
    6,592
    Joined:
    May 2, 2007
    Location:
    An Australian in Singapore.
    I did a lot of testing at 44.1khz versus 96khz- yes when dealing with high end converters, monitoring and a well designed room(which I have) there is a small difference.
    For most people with 'prosumer' converters and monitoring, working in bedrooms there is no way of hearing it.

    Most of my work involves 40-60 tracks of audio with a lot of ITB processing.
    It isn't practical for me to try to run those sessions at 96k for such a small return in performance.

    I mostly work in 44.1 or 48khz and don't see it changing for a while.
     
  12. kp8

    kp8 Friend of Leo's

    Posts:
    4,981
    Joined:
    Aug 22, 2004
    Location:
    Seoul, Korea
    Each bit you add doubles the resolution. So 24 bits is going to be way way bettererer than 16 and really does give a lot more room for volume and more resolution on the quiet stuff. A big deal imo.

    Meh. I disagree. The difference between 44.1 and 48 is tiny in terms of sample resolution and staying at 44.1 avoids resampling. I think you are underestimating what happens when you resample a waveform and over estimating (wildly) the benefit of an extra 3900 samples. Whatever works for you but I have done some digging in this area (i'll leave it at that) and very few packages have good resample routines and only the very best ones come close to clean resampling. You would be shocked how dirty many graphs of (even high end) resample routines are. There used to be a site that hosted data on this but it seems to have disappeared. Oh how handy it would be now to see those nasty filters now.

    It isn't about having the juice to resample it is about how good your resample algorithm is, how good the filters are, etc. Most commercial software packages perform poorly in this area and just use crudely implemented band-limited interpolation. In most cases you are adding more artifacts than you would be reducing my such a tiny bump in sample rate.

    YMMV but total waste imo.

    -kp

    ps. one of the best resampling packages out there is sox, which is open source. It is pretty slow (though not such a problem these days) but really worth it to use a good algorithm. It isn't a sexy thing feature to add so lots of fancy packages have chugging along with their same zippy but crude resampling routines from way back when. There were some high end plug ins that also do a good job but i have yet to find one that bests sox.
     
  13. gtrguru

    gtrguru Friend of Leo's

    Posts:
    2,352
    Joined:
    Apr 29, 2011
    Location:
    Detroit Metro
    For me it depends on the dithering I have at my disposal. My go to on most home studios is 24/44.1. I have personally experienced sound degredation at a studio going from 48k to 44.1 and one bad experience is enough. Depending on the equipment most pro studios can do 88.2 without any loss in quality. IMO!
     
  14. fezz parka

    fezz parka ---------------------------

    Posts:
    13,841
    Joined:
    Jan 22, 2004
    Location:
    Del Floria's Tailor Shop
    24/44.1. If you're doing orchestral stuff or acoustic type music, then 88.2.
     
  15. yardleyboy

    yardleyboy Tele-Meister

    Posts:
    426
    Joined:
    Jan 23, 2011
    Location:
    UK
    Good for you

    But I think you are kidding yourself
    Unless of course you have perfect hearing, are under 25 years old, have a perfectly prepared acoustic environment' hugely expensive mics, and a monitor system that cost you $20k. Then .... I still think you are kidding yourself.

    24bit is the key part - gives very good headroom
    24 bit, 44.1 is more than good enough for anybody who records at home or even as a very serious hobby

    48 is traditional in video/film as a standard
    Some high end classical recordists now use 96kz sampling
     
  16. yardleyboy

    yardleyboy Tele-Meister

    Posts:
    426
    Joined:
    Jan 23, 2011
    Location:
    UK
    You may have done, but it wasn't cos of the sample rate change
    Impossible to hear the difference unless the conversion was done incorrectly
     
  17. yardleyboy

    yardleyboy Tele-Meister

    Posts:
    426
    Joined:
    Jan 23, 2011
    Location:
    UK

    All +1
     
  18. octatonic

    octatonic Poster Extraordinaire

    Posts:
    6,592
    Joined:
    May 2, 2007
    Location:
    An Australian in Singapore.
    Just to clarify this.
    Resolution is a term usually applied to sample rate, not bit depth.

    Bit depth is dynamic range- this isn't resolution.
    16 bit gives you a 96db dynamic range.
    24 bit gives you a 114db dynamic range.
    Dynamic range is simply the amount that a signal can reduce form 0db before it disappears into the noise floor of a converter.

    Curiously much commercial pop music uses around 6-20db of dynamic range.
    Really to hear modern converters use the full dynamic range you need to go to quality classical recordings or the Chesky label.
     
  19. Larry F

    Larry F Doctor of Teleocity Vendor Member

    Posts:
    16,630
    Joined:
    Nov 5, 2006
    Location:
    Iowa City, IA
    I'm 24/44.1.

    I do have to register another variation of an old gripe of mine. The original gripe was the subject of one of my threads last week. When people are talking about this amp vs. that amp, or speaker vs. speaker, or even cable vs. cable (I list this last one only for sport), far too many claim that the differences or nuances will be lost in a gig, due to the less than ideal listening environment. At the core of this gripe is that these claims are presented as thought experiments which supposedly would have the same outcome if they were done as actual experiments. I just don't like those kinds of though experiments in general, as they purport to tell me that I am not hearing correctly if I claim to be discerning in some way. According to the thought experiment, research would, if it were actually done, show that my claims are not borne out with listeners in a real world gig situation.

    A variation of this is being done in some posts above, where someone says you would never hear the difference between 44.1 and 48 under normal circumstances. Maybe this is correct, but how do I evaluate that? I'm not saying that I am pro 48--as I mentioned above, I use 44.1. I just get my guard up when people tell me how normal people are supposed to hear in normal situations. The only time the high-end interpretation works, we are told, is only in extreme situations with ideal listeners, environments, and gear. Again, maybe these people are correct, but isn't there a better way to make that case than to conduct a faux experiment.

    In other kinds of listening experiments that musicians talk about it, other claims are made. A well-known claim among composers a generation ago was that one cannot distinguish between a serial composition and one created through random procedures. Foolishly, or maybe just naively optimistic, a student of mine once sought to demonstrate this by playing an excerpt from the music of John Cage and another by Milton Babbitt. Of course, the differences are night and day, to large extent because serial compositions have a lot of features that sort of go along with the serial technique, but are not an intrinsic part of it. Ditto with Cage.

    For sound, in my experience, if I cannot discern a difference in quality between two operations or standards, that does not mean that I won't hear a difference farther down the road, as when, for example, a sound is pitch shifted again and again. By, say, the 5th transformation, will a 44.1 sound be perceived as identical to a 48 sound? In other words, you may not hear a difference between two sounds presented in a direct manner, but that doesn't mean that drastic differences won't emerge a little later down the line of processing and other transformations.

    For me, this is a very real problem, as much of my creative work involves sounds that have been processed again and again. Now, I must admit that I am lazy and that I do not like to change my standards and methods, to my detriment as a composer, for sure. Like anyone, I can talk myself out of problems like this by claiming that any sound that is processed a number of times will have noticeable differences in quality compared to the original sound. In my case, then, I would say "don't process a sound so many times," instead of "use 192 if you want to process many times."

    This may look like gibberish, but I'm trying to make concrete something that is felt as much as it is heard. Also, I'm well aware that I am conducting my own thought experiments, so guilty as charged.
     
  20. 64Strat

    64Strat Friend of Leo's Vendor Member

    Posts:
    4,756
    Joined:
    Oct 6, 2003
    In my system, it works for me.
     
IMPORTANT: Treat everyone here with respect, no matter how difficult!
No sex, drug, political, religion or hate discussion permitted here.


  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.