Needledroppers: Why bother using noise shaped dither?

Discussion in 'Audio Hardware' started by gloomrider, Jan 29, 2015.

Thread Status:
Not open for further replies.
  1. gloomrider

    gloomrider Well-Known Member Thread Starter

    Location:
    Hollywood, CA, USA
    My inspiration for this thread is a conversation that @Grant and I had about this a bit ago.

    I'm completely agnostic about this, so I don't think there's such a thing as a "correct" answer.

    My thesis: The noise floor of a needledrop is far higher than what is achieved with non-noise-shaped (a.k.a. "flat" dither). What benefit (if any) is achieved by using noise shaped dither?

    Thoughts?
     
  2. Grant

    Grant Life is a rock, but the radio rolled me!

    Well, I just had another discussion with another member about my settings in MBIT+, and, since then, I have totally changed my preferences. I am using some presets in iZotope that get me closer to there I want to be. I first used my new dither settings on a hi-rez files without sample-rating it down to 44.1 first to get a fairer assessment of the dither. Then, I did a test with SRC first, then the new dither/noise shaping setting, and it's closer. I'll post it later. My pizza is about ready.


    OK, I just took it out of the oven. MMmmmm! Anyway, to answer your question from my opinion: Flat dither does leave a higher noise floor that I don't like. So, that is why I like noise shaping. All it does is move the dither noise to less noticeable areas of the frequency spectrum that you can specify. The trick is to not adversely alter how it makes the rest of your music sound. Too much moved up to the upper region can result in uncrisp highs. Too much at the lower side, and it can muddy up the bass. Too much in the midrange will be heard and mess with that. Too much dither can be audible and counter-productive. Too little will result in more graininess. That's why you gotta do your own listening tests.
     
    Last edited: Jan 29, 2015
  3. gloomrider

    gloomrider Well-Known Member Thread Starter

    Location:
    Hollywood, CA, USA
    Thanks for this. Any chance you could post a snippets of your dither prefs vs flat but label the files like "A" and "B" or something? I've done ABX of flat vs. shaped for my needledrops and just can't tell the difference.

    Is the pizza pepperoni? :drool:
     
    Grant likes this.
  4. Grant

    Grant Life is a rock, but the radio rolled me!

    While you were salivating, I just now added more to my post.
     
  5. Grant

    Grant Life is a rock, but the radio rolled me!

    Keep watching this thread. I may get that done later.

    Is the pizza pepperoni? :drool:[/QUOTE]
    :agree:
     
  6. gloomrider

    gloomrider Well-Known Member Thread Starter

    Location:
    Hollywood, CA, USA
    Can you characterize the vinyl noise floor and the dither noise floor in dBFS? I'm still not convinced it's possible to detect the dither noise above the record noise. But maybe your vinyl rig has a super low noise floor?
     
  7. Grant

    Grant Life is a rock, but the radio rolled me!

    Why are you thinking about the vinyl noise? You don't dither to lessen vinyl noise. You use it to reduce quantization noise. You use noise shaping to tame the dither.
     
  8. gloomrider

    gloomrider Well-Known Member Thread Starter

    Location:
    Hollywood, CA, USA
    I thought noise shaping was used to lower the noise floor in the audio spectrum where it is most noticeable and move it to frequencies where it is less noticeable.

    Flat dither makes a uniform noise floor. I'm thinking the level of that (in dBFS) is less than the noise floor of the vinyl. What am I missing?
     
  9. Grant

    Grant Life is a rock, but the radio rolled me!

    Not the noise floor, the dither noise itself. Dither is not meant to do anything to the noise already inherent in the recorded sound. You want a cleaner vinyl sound? get cleaner vinyl, or use noise reduction.

    The part about vinyl noise being a factor in any of it. One uses dither to reduce or eliminate quantization noise caused by truncation.

    Let's get away from vinyl for a moment. Let's consider, for example, a master recording on 1/4 tape at 30 i.p.s., no noise reduction used. The tape still has some tape hiss in the signal. You transfer that tape to 24-bit. You use dither to get it to a CD. If you don't use noise shaping (flat dither), that dither noise will be on top of whatever is on the tape. If you used noise shaping, all that is supposed to do is move your dither noise to less perceptible/more agreeable areas of that audio. It should do nothing to that tape hiss already in the signal. This same thing goes for vinyl noise.

    Now, some flavors of noise shapes can move the dither noise around to make it appear that the noise floor is lower than what it really is, if you move that dither noise to the right places. That noise floor, and all that dither, is still there.

    The reason flat dither is favored today, is because when you process a dithered file, you risk increased noise. You, as the engineer, do not know, or have control over whatever that end user does, including online vendors like iTunes and Amazon. And, since further processing an already dithered file at 16-bit can really ruin the sound, it is safer if simple, flat dither is applied. It doesn't sound all that great, but it's safer for the end-listener. Besides, even though, today, noise shaping can be virtually transparent, many engineers already have a negative experience burned into their heads. The older dither flavors of the month that were used back in the 90s changed the timbre of the sound.

    Some may ask, why bother with 20, 22, or 24-bit, anyway? Well, if you are going to process in any way, you have to do it in a high bit-depth, or you will risk cumulative dither, or quantization noise, probably from wonky plug-ins. And, many people, like me, feel that 16-bits do NOT capture all of the sound on a tape recording. A 16-bit recording done at 24-bit, then rendered to 16-bit, will still sound superior to one simply encoded at 16-bit.
     
    Last edited: Jan 29, 2015
  10. gloomrider

    gloomrider Well-Known Member Thread Starter

    Location:
    Hollywood, CA, USA
    Ok, I'm following you I think. The noise is additive. So if vinyl noise = X and dither noise = Y, the total noise is X+Y.

    I use 32 bit float in the DAW, then finish to 24 bits for sample rates above 44.1kHz and 16 bits for 44.1kHz. While you don't have to sell me on higher sample rates, they do make bigger files and I'm finding I'm having to make one set of files for home listening and another for portable (DAP) use. When 512GB DAPs become ubiquitous, I can just finish in 24/96 :)
     
  11. Grant

    Grant Life is a rock, but the radio rolled me!

    Why can't you do that now? Just buy an extra hard drive to store you hi-rez on. I'm still using a duo-core processor and 8GB RAM. I can handle 24/06 just fine.
     
  12. gloomrider

    gloomrider Well-Known Member Thread Starter

    Location:
    Hollywood, CA, USA
    You misunderstand. I make two files now. But I would like to be able to put the 24/96 files on the DAP and not have to make a second 16/44 file.
     
  13. Grant

    Grant Life is a rock, but the radio rolled me!

    That's what I mean. Why can't you play the hi-rez on the DAP?
     
  14. gloomrider

    gloomrider Well-Known Member Thread Starter

    Location:
    Hollywood, CA, USA
    I can play it just fine. But I would need a lot more GB on the DAP to put all needledrops on it in 24/96. Like I said, we need a 512+ GB DAP!
     
  15. Kyhl

    Kyhl On break

    Location:
    Savage
    Maybe off topic, but why use a DAP? Are you bringing the lossless version on the road?
     
Thread Status:
Not open for further replies.

Share This Page

molar-endocrine