Best way to convert 24/96 to 16/44.1?

Discussion in 'Audio Hardware' started by Zoo Station, Sep 5, 2006.

Thread Status:
Not open for further replies.
  1. Zoo Station

    Zoo Station Active Member Thread Starter

    Rochester, NY
    Anyone got an acceptable procedure to convert a 24/96 needle-drop to 16 bit 44.1 for CD burning purposes? Will the conversion "wreck" the quality? Thanks.
  2. Grant

    Grant A Musical Free-Spirit

    No matter what you do, any conversion will chang or damage the sound to some degree. The idea is to make that conversion as transparent as possible. For that, you need a good program and/or hardware that let's you tweak the settings.

    Now, if your plan is to go to redbook, don't record at 96k, record at 88.2, if you can. Why? Do the math, 88.2 divides down to 44.1, and will sound cleaner because of it. The reason 96k is so popular is because it is the industry standard. 48k is also the industry standard, so 48x2=96. It's just like 192k is more popular than 176.4k, wich is just 88.2x2.

    Dither is necassary but evil. It is literally analog noise added to a digital word when converting to 16-bit to reduce the effects of truncation. Use too much, the music sounds noisy and dull. Use too little, the music sounds brittle and noisy. Some maintain that dither is only important at low levels, like on fade-outs, but I say it affects the sound overall, and that is very audible.

    You should always convert the sample rate before you dither down to 16-bit. Most software programs will do the conversions in that order if the user chooses to do it that way. And, your files should be only dithered once. And, as you should know, dither should be the very last thing you do.

    BTW, noise shaping can be employed to reduce the negative effects of dither. Noise shaping doesn't remove the dither, it simply moves the dither noise around to parts of the spectrum that may be less sensitive to the human ear. Again, the best thing is to have a program that lets yoy tweak the settings to taste and function.
  3. Zoo Station

    Zoo Station Active Member Thread Starter

    Rochester, NY
    Thanks for the reply. Pretty much what I thought. My friend did it cause he wanted a hi-rez version, then later he wanted to burn it to redbook without doing another needle drop.
  4. Grant

    Grant A Musical Free-Spirit

    Even if you want 96k, and you need to go to redbook, what you can do is record at 88.2 and when you need 96k, just upsample. 88.2k and 96k are so close, there's really no loss in quality.
  5. GT40sc

    GT40sc Active Member

    Grant is right when he says, "any conversion will change or damage the sound to some degree."

    Yet we have to accept that change, if we want to listen on a standard 16-bit/44.1 CD...the trick is to minimize the damage as much as possible...

    So try it every way you can...use all the programs you have, burn a disc each way, and see which one sounds the best.

    You might be surprised at what you find. On my system, I can hear a difference if I apply dither at the "bounce to disc" stage, or if I put the hi-rez file into the CD-burning program, and apply the dither there instead...

    One more secret from the pro-audio world. (Don't laugh)...try an analog dub. Hook up your best cables, play the hi-rez file, and make a straight 16/44.1 copy onto another deck.

    It won't sound as good as the 24/96 file, but the damage might be less than you think...

    best of luck,

  6. Metoo

    Metoo Forum Hall Of Fame

    Spain (EU)
    Grant, why not do it the other way around, record at 96 and then downsample to 88.2 if you are going to 44.1. In this way you make sure that your HiRez version is as clean as possible. As to what I suggest here, since you lose a lot down to 44.1 downsampling first to 88.2 should not be as noticeable.
  7. Grant

    Grant A Musical Free-Spirit

    In my experience it doesn't quite work that way. It's all in the math.

    There is a small difference between 96k. You can probably hear the damage doing it your way. I can, anyway. Since they are so close, you can upsample with no loss in quality. Your 88.2 will sound like 96k anyway.

    See, the indistry has everyone thinking that 96k is the end-all, be-all. Andm that is fine. But, like I said before, 48k is the industry standard and does not divide evenly into 44.1. Same with 96k. Many pros do realize that if you go to CD, 88.2 is the better option. Forum member Jamie uses 88.2.

    TONEPUB Forum Resident

    Portland, Oregon
    Im not sure about this yet, but I recently picked up a Benchmark ADC 1
    that I've been too busy to get to. If I read the manual correctly, it provides outputs at 192k and also at 44.1 simultaneously. This could be very helpful.

    more notes to come from the hi rez future....
  9. strat95

    strat95 Active Member

    Try downsampling from 96 to 48 and then from 48 to 44.1. This simplifies the math the most (damage will still be there but minimized when compared to a downsampling from 96 directly to 44.1).

  10. bluemt

    bluemt Member

    Lincoln, MA 01773
    You should consider sending it out analog and capturing it at 16 bit 44.1. the da/ad might be less damaging than the sample rate conversion depending on the gear you have at your disposal.
  11. keoki82

    keoki82 Active Member

    I don't mean to hijack this thread, but my question has some relation to the original post. When ripping CDs using EAC, is there a way to rip to 24-bit instead of 16-bit? The reason I ask is because when ripping to wave I usually like to normalize to average RMS power to make my tunes "sound" the same level (when making a CD-R compilation). If I am doing this at 16-bit, there's probably more of a loss in sound quality than there would be at 24-bit, correct? So would it be good practice to find a program that would rip directly to 24-bit, or should I simply upgrade the bit depth of the waves to 24-bit before normalization, and then dither back down to 16-bit for a compilation? - Additionally, I'd like to get into storing PCM data as lossless. I've worked with both ALAC (Apple) and Sony Perfect Clarity in the past, but these formats are too proprietary for my liking. FLAC seems to be highly favoured. Does FLAC support 24-bit files?
  12. acjetnut

    acjetnut Well-Known Member

    The numbers stored on a CD are in the 16 bit realm, so putting them into 24 bit "boxes" doesn't help. Any program that does the normalization should work internally in at least 24 bit.

    As far as recording goes, it is very common for mastering engineers to have an analogue chain.
  13. Metoo

    Metoo Forum Hall Of Fame

    Spain (EU)
    Sorry to digress with you here, Grant, but 88.2 -albeit close- is not 96/24 and if you upsample it you will only get 88.2 disguised as 96khz.

    I do agree, though, that 96/24 is not the end-all, it is just an agreed on spec for DVD. If I had my choice I'd choose 192/24 all the time, even if the differences are more subtle the more you go up in resolution.

    If I remember well, I must have been one of the first to mention on this forum that using a frequency that is a multiple of the target downsampled one to achieve better results. If in doubt, do a search.

    I am not saying that you should not use 88.2 i f you later intend to downsample to 44.1. What I mean is that, IMHO, it is better to start at 96/24 (then you already have a DVD format compatible version), then downsample to 88.2 before going to 44.1, instead of what you mention of recording first at 88.2 and upsampling that to 96/24 as you cannot add what is not there before, but the loss of information is so great when you go to low resolutions such as 44.1 that going 96 > 88.2 > 44.1 shouldn't be noticeable.

    Maybe the differences here are due to the fact that, IIRC, you tend to gravitate towards 44.1 in your projects and I tend to go for 96/24 and 192/24 in mine. The reason for this is that, IIRC again, you tend to go for the CD format and I tend to use DVD as my target one.
  14. Metoo

    Metoo Forum Hall Of Fame

    Spain (EU)
    Sorry, Strat, but that won't work. 48 is not a multiple of 44.1 and you'll get all the decimals from the operation and, thus, a more altered sound. 88.2 > 44.1 is the way to go here. In that both Grant and I agree. The only thing we do not coincide in is what you might start from depending on the project.

    He says, start at 88. and, if you need 96, upsample. I say, if you are ever going to need a higher sampling frequency start at it and go down from there. I imagine that for some people here, not meaning necessarily you, especially beginners, this might seem much like splitting hairs.
Thread Status:
Not open for further replies.

Share This Page