1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

Some audio quality questions..

Discussion in 'Recording' started by cotenyc, Mar 27, 2007.

  1. cotenyc

    cotenyc Active Member

    Hey everyone!

    I'm recording in Sonar, now at 24/48, but some of my projects are now mixed 16 and 24bit in one project. When I export the final .wav mix, am I sacrificing quality by exporting to a 24bit file if some audio is originally recorded at 16bit?
    What happens to the 16 bit audio, is it upsampled?

    Also, If I have my 24 bit tracks ready to burn on CD, I understand there is a conversion process in which files do get burned at 16bit.. Is Nero good for this, or does the program not even matter?

    thank you all
     
  2. Boswell

    Boswell Moderator Distinguished Member

    You are confusing wordlength with sample rate. The 16-bit data is effectively padded out with 8 zeroes to become 24-bit data and then mixed with the real 24-bit data. This doesn't alter the quality of the 16-bit data. I said "effectively", because it's actually more complicated than that. All the processing is done in floating-point arithmetic, so there are fixed-to-floating and floating-to-fixed conversions being performed at the time of importing and exporting the data. This has nothing to do with upsampling or downsampling, but see below.

    No, it does matter and Nero is not good for this. You need to convert your final mix to be 16-bit 44.1KHz before it can be written to CD. Nero is fine for doing the actual CD writing, but don't use it for making any changes to your audio data.

    You should use your audio processing software (Sonar or whatever) to wite the final mix as a .wav file, but set it to dither the data to 16 bits.

    This leaves you with the other problem of having recorded your data at 48KHz and needing to write the CD data file at 44.1KHz. This is where the upsampling/downsampling comes in, with processing called (naturally enough) sample rate conversion (SRC). Again, you will have to call on your audio processing software to perform this operation. However, I usually advise people to record at 44.1KHz (or 88.2/176.4) if the result is to end up on CD, just because I think that, no matter how good the SRC process is, it produces subtle artifacts in data derived from 48 KHz recordings.
     
  3. cotenyc

    cotenyc Active Member

    thanks boswell,

    Am i better off recording at 44.1 or 88.2... huge quality difference? worth the extra space?
    Thanks.
     
  4. RemyRAD

    RemyRAD Well-Known Member

    cotenyc you might just want to think about staying at 16-bit 44.1. There really isn't any reason not to work in that format. Then nothing has to be converted. Nothing has to be translated. Nothing has to be altered. Everything is compatible.

    I've worked this way since it was invented. I have no problems with 96 DB of dynamic range. I don't like but accept the fact that brick wall filtering has to be used to be able to place audio on a audio CD, at 44.1kHz.

    Sure, 88.2-96-192kHz and 24 to 32-bit sounds better than 16-bit 44.1kHz but then, do you plan to release in a high-definition format? Personally, I think all PCM sounds like crap in comparison to DSD and you can buy a brand-new DSD recorder from Korg, starting at $750 US. It doesn't get better than that. I just can't imagine, $750 but it's true. Of course that is strictly a 2 track portable recorder with built-in rechargeable battery but it is DSD, 1 bit sampled at 2.53MHz, with an effective audio frequency response to 100kHz.

    I know this is going to blow everybody's mind!
    Ms. Remy Ann David
     
  5. cotenyc

    cotenyc Active Member

    hey Remy,
    I hear ya completely, I used to record everything at 16/44.1 and it did sound great, but recently I bought a mackie onyx 400f, which is locked at 24 bit, but the sample rates are adjustable. I figure that since I'm already recordng at 24 bit, I may as well up the sample rate as well.
     
  6. IIRs

    IIRs Well-Known Member

    So lets get this straight Remy: you claim to be able to hear the jitter of the cpu clock when summing tracks ITB, but you can't hear the difference between 16 and 24 bit audio???

    :?

    This is really BAD advice. Nothing has to be converted?? I take it your DAW software performes all calculations at 16 bit resolution then? No wonder you think digital summing sounds bad! :lol:

    Actually your DAW software almost certainly performs all calculations (including simple gain changes) at 32 or 64 bit floating point resolution, which means your word-length will need to change whether you record 16 bit files or 24 bit files. Choosing 16 bits simply means reduced dynamic range and resolution: if you leave 6dB of headroom you will effectively end up with only a 15 bit recording, and if you record a very dynamic performance which later needs plug-in compression you may find the quiet passages have only 12 bits of resolution, or less.

    Yes 16 bit audio technically has a dynamic range of 96 dB, but unless you start off with a much higher resolution and then dither it down at the final stage the bottom few bits will just be distorted rubbish.

    16 bit audio is a consumer format IMO: you should no more use that for audiophile recordings than a cassette multi-tracker.
     
  7. Boswell

    Boswell Moderator Distinguished Member

    cotenyc - if you record using your 400F at 44.1KHz, you will get great recordings that will accurately represent your skill as a sound engineer. Don't get too bogged down in wordlength arguments - all you have to do is to set the program to dither at the 16-bit level when writing the final mix.
     
  8. BobRogers

    BobRogers Well-Known Member

    I have pretty much settled on recording at 44.1/24. I figure that the 24 bits is worth the space on headroom considerations alone. I suppose if I had more experience I could record at 16 and use a lot of bits without ever clipping, but if I'm using 24 bit words I can use 16 bits to sample an average volume level and still have 8 bits left to handle peaks.
     
  9. cotenyc

    cotenyc Active Member

    gotcha bos, thanks
     
  10. RemyRAD

    RemyRAD Well-Known Member

    I'm not saying that I can't hear the difference. When it comes to the differences, PCM sound sucks regardless of the sample rate or bit depth. Nuance differences are just nuance differences when it comes to PCM. I don't really think it warrants the cost difference, especially for those just starting out. Nobody complains about my 16-bit recordings at 44.1kHz because it really doesn't make any difference. It's the engineering technique above all else. Not equipment blah blah.

    I can hear a huge difference between any PCM format and DSD, which is the basis for the SACD format of super audio CDs. DVD-A, is a nuance difference from 16bit, 44.1kHz and another failed and flawed format. Sure, getting rid of brick wall filters does a lot to improve the sound but what about CDs? You're back to go again. That sort of like putting in "AV gas" into your Toyota Corolla. We worked in 16-bit 44.1kHz PCM because it was the first back in the early 1980s, then it became inexpensive, convenient, plentiful digital recording system and the basis of audio CDs along with being a fairly well-established standard from the get-go for a crappy digital audio format. It's still sort of like owning a Ferrari. You're still only allowed to drive 55 mph on most of the nation's highways. When you don't drive under that well-established standard, conversions, translations and the risk of incompatibility occurs, not to mention catastrophic failures. That's why I really don't care if I make a recording with a bag full of SM57's and a Mackie mixer or, my 36 input Vintage Neve and a dozen or two Neumann's. My recordings come out beautifully crappy the way everybody likes them. Fat, punchy, with beautiful stereo and imaging and still at 16 bits.

    Good engineering is good engineering and my stuff at 16 bits, along with other well-known and successful engineers did quite fine throughout the eighties and nineties in that "less than audiophile" format. They still won awards. And you can too. If your system defaults to 24-bit/32-bit float and you have the hard disk space and CPU power, nothing wrong with working in 24-bit, etc.. It still has to dither down to 16, to stick on a CD and if you think CDs sound that bad, why would anybody want an i-Pod or ever download an MP3?

    No Ferrari and no DSD......yet
    Ms. Remy Ann David
     
  11. cotenyc

    cotenyc Active Member

    Guys,
    I am bit confused, after reading some more I keep seeing many people say how one should raise the rate higher than 44.1..?
    I am recording at 24 bit and it does sound good, but am I missing something by not using 96 or 88 or even 48 or is it so negligible that it doesn't even matter.
    Its not the hard drive space that is a factor, I have a decent amount and I want to have very great quality recordings.

    Thanks again!
     
  12. BobRogers

    BobRogers Well-Known Member

    You have all the equipment to answer this for yourself. (1) Record something reproducible at 44,48,88,96. (2) Dither them all down and burn to disk to 44. (3) Put the CD on random select. Can you tell which recording is playing?

    I did it. I'm recording at 44. Might be a comment on my ears, but there it is.
     
  13. RemyRAD

    RemyRAD Well-Known Member

    XACTLY! Thank you Bob! Us folks in Virginia have to stick together.

    Sorry Jeremy
    Ms. Remy Ann David
     
  14. dementedchord

    dementedchord Well-Known Member

    a minor quibble... it seems some vst's sound better and a lot of people get lower latency while running at higher rates... that's why i do it...
     

Share This Page