Dither without noise shaping/flat dither?

[Maraxvx]

I notice that a lot of professional records don't seem to have that high frequency noise shaping that you normally get from dithering down.

My question is can you dither WITHOUT noise shaping?

For instance in Ozone MBIT+, 16, with no shaping....
Or L2 with type 1, 16, and shaping set to none?

Is this what "flat dither" means? Normally I add noise shaping, but I don't know if it is right for my music (that's a little harder edged).

[Jon Hodgson]

Dither and noise shaping are different processes, although they are combined into a single operation (they both affect the quantizer, with dither being placed before it, and the noise shaping in a feedback loop), so yes you can do one without the other.

Note that noise shaping is separate from choosing the shape of your dither (TPDF etc).

[HomeProducer]

Dithering is adding noise to randomize the quantization for small signals and avoid in this way some sort of parasitic tones. Flat dithering is adding flat noise in the whole spectrum. Noise shaping tries to add it outside the hearing range.

[blusound]

I know it should be the contrary, but in my experience noise shaped dither is more audible then TDPF dither.
Best regards
Leo

[editronmaximon]

Originally Posted by blusound

I know it should be the contrary, but in my experience noise shaped dither is more audible then TDPF dither.
Best regards
Leo

Probably less harmful to humans, too.

Noise shaped dither is just asking for trouble. Who really thinks that the product is always going to be played back on a flat system? No, its often played back with the treble boosted.

[HomeProducer]

Originally Posted by editronmaximon

Probably less harmful to humans, too.

Noise shaped dither is just asking for trouble. Who really thinks that the product is always going to be played back on a flat system? No, its often played back with the treble boosted.

beyond 20 kHz where most systems have strong roll-off anyway

[Maraxvx]

so, would i get flat dithering by putting in the settings on the l2 idr or ozone type 1, 16 bits, but with no shaping? or is it through one of the pow-r dithers?

i think i understand what you guys are saying. flat dither adds noise to the entire spectrum while noise shaping moves the noise to the higher frequencies.

i have never tried flat dither before. i was always under the impression that noise shaping and dither were supposed to go together at the final stage of bouncing down. i am going to try idr/ozone mbit+ without noise shaping asap and see what the results will be.

[studioland]

Dither is random noise that is added to a signal BEFORE truncation at the LSB of the destination bit depth.It is independent of (random or not correlated to) the original signal. What dither does is prevent the signal from ever falling below the level of the LSB, so a D/A converter never turns off and therefore truncation distortion is avoided altogether. This is not magic. There is still a loss of dynamic range and a tiny bit of noise added at the LSB but there is no longer the edgy distortion.
But wait, there's more: There is a further benefit in that signals that might fall just below the threshold of the LSB are mixed with the dither and now might rise to just above the level of the LSB. In effect, you might be able to realize greater than 16-bit performance with a properly dithered signal!


"Flat" dither is similar to pink noise (equal energy per octave) the most common type being TPDF (triangular probability density function).

"Noise-shaped" dither, which means that more of the dither energy resides in the high-frequency area just below the Nyquist frequency. The theory behind noise shaping is that the dither itself is less likely to be noticeable itself and it is less likely to mask low-level information in the band where our hearing is most sensitive




If you have ozone I suggest you play around with different settings and most of all read the manual it is explicated very well and in details. It will help you understand better dithering and which one you may want to choose.
AS with anything else I also suggest you experiment and with a little bit of practice ...

Only dither at final stage and check out how the soft you're using appklies this because softwares react differently
Only dither if you master the music not if you're just mixing.

[editronmaximon]

Originally Posted by HomeProducer

beyond 20 kHz where most systems have strong roll-off anyway

I don't think so. Maybe UV22, but that's not really dither, is it? Flat TPDF is recommended unless you are absolutely sure that the file will not undergo any further processing, and you can never really be sure of that.

[editronmaximon]

Originally Posted by studioland

Dither is random noise that is added to a signal BEFORE truncation at the LSB of the destination bit depth.
[B]

TPDF at 2LSB of destination bit depth is what is deemed necessary to provide the claimed effect.

[blusound]

Originally Posted by studioland

"Flat" dither is similar to pink noise (equal energy per octave) the most common type being TPDF (triangular probability density function).

Pink noise is a shaped noise. Flat dither is similar to white noise: totally random probability.
Best regards
Leo

[Jon Hodgson]

Originally Posted by studioland

"Noise-shaped" dither, which means that more of the dither energy resides in the high-frequency area just below the Nyquist frequency. The theory behind noise shaping is that the dither itself is less likely to be noticeable itself and it is less likely to mask low-level information in the band where our hearing is most sensitive

Shaped dither is not noise shaping.

There are a number of shapes of dither noise which attempt to achieve a less noticable addition of noise while still correctly dithering the signal. This is what you are choosing when you select TPDF, RPDF etc. It's usually a fixed system, the shape of distribution of that dither remains constant regardless of the signal content.

Dither cannot give "better than X-bit performance", what it gives is correct x-bit performance, in other words your original signal plus broadband noise of some description.

Noise shaping on the other hand is a negative feedback system, the error coming out of the quantizer (the difference between the newly quantized signal and the original) is fed back to the input in such a way that it moves the errors around, usually up the frequency range.

Noise shaping can achieve "better than X-bit performance" in a given frequency range (in exchange for worse performance in other ranges), just listen to DSD if you doiubt it, that's a 1 bit quantizer with a lot of noise shaping to move the errors above 20kHz.

You can actually noise shape without dither, though I wouldn't recommend it.

[Alexey Lukin]

Originally Posted by Mara***

so, would i get flat dithering by putting in the settings on the l2 idr or ozone type 1, 16 bits, but with no shaping?

Yes.

[HomeProducer]

Originally Posted by editronmaximon

I don't think so. Maybe UV22, but that's not really dither, is it? Flat TPDF is recommended unless you are absolutely sure that the file will not undergo any further processing, and you can never really be sure of that.

UV22: It is dither, why not? Has the same impact.

I agree though that pink noise is better suited to keep a constant SNR in every octave.

[Maraxvx]

Yeah, I understand perfectly well about dithering only once. Usually I dither with a loudness maximizer right before I bounce the track down to 16 bits. However, with that said I am still confused about noise shaping. Do I want or need noise shaping? Does it sound better than dither without noise shaping?

Doesn't noise shaping add high frequency noise that is capable of distorting certain playback systems? If that's the case, then I would rather just go with flat dither noise. I don't mind a little subtle noise in my recordings.

What kind of dither would you want for a harder edge style of music like heavy metal or punk rock? What kind of dither would you want for orchestra/soundtrack CDs with a wide dynamic range?

[dcollins]

Originally Posted by Mara***

Yeah, I understand perfectly well about dithering only once.

We d*ther every time the word-length is reduced. From 48 bit to 24 or from 24 to 16 or from 16 to 8. It's not a one-time thing.

Usually I dither with a loudness maximizer right before I bounce the track down to 16 bits. However, with that said I am still confused about noise shaping. Do I want or need noise shaping? Does it sound better than dither without noise shaping?

Leave the noise-shaping off, and personally I would focus my energy (Har!) elsewhere.

Doesn't noise shaping add high frequency noise that is capable of distorting certain playback systems? If that's the case, then I would rather just go with flat dither noise.

The level is too low to cause a problem like that.

I don't mind a little subtle noise in my recordings.

Yes.

Before digital, if you were to ask an engineer how much time he spent worrying about noise below -90, he would think you were crazy

What kind of dither would you want for a harder edge style of music like heavy metal or punk rock? What kind of dither would you want for orchestra/soundtrack CDs with a wide dynamic range?

It doesn't work like that. Just use the flat or HPDF and move on.


DC

[Maraxvx]

I'm just going to use the noise shaping for waves and ozone. After reading the manuals for both it seems like this is the best option. It doesn't seem to be that big of a deal after all.... Thanks!

[Jon Hodgson]

Originally Posted by dcollins

Before digital, if you were to ask an engineer how much time he spent worrying about noise below -90, he would think you were crazy

It's nuts. You read interviews about various famous (and great sounding) albums and they're chopping up the tape, copying it, putting stuff through various noisy units, etc etc, all with a medium that has severe limitations in terms of noise, distortion and crosstalk, and nobody seemed to worry about it and certainly the listeners aren't complaining.

Now you get twenty page threads about whether method A or method B is better because people are scared that someone's going to notice the difference 130 dB down. People obsess about putting the signal through one process that has noise and distortion figures you could only dream of in the analogue world, whereas those guys wouldn't have thought twice about putting it through several far more destructive processes.

Educate yourselves by all means, but get it in perspective and get on with your art (that way I have something nice to listen to while I worry about the little numbers in my work).

If you think it may go through another process, then TPDF dither with no noise shaping is probably a good bet, since it results in pretty well white noise. If it's for final delivery, just listen, pick one you like, and stop worrying about it.

[Maraxvx]

Sorry... but.

TPDF dither is Type 2 in Ozone, right?

So, I would select Type 2 to dither with no shaping?

[studioland]

Just experience ... every dither option sound different so if you are happy with your master before dithering you should take 2-3 minutes to decide and test which one you would like to use on the given material. It does affect the sound though of course this is a very little step in the whole process

[jayfrigo]

Originally Posted by studioland

Just experience ... every dither option sound different so if you are happy with your master before dithering you should take 2-3 minutes to decide and test which one you would like to use on the given material. It does affect the sound though of course this is a very little step in the whole process

If, as you say, you like your sound before dithering, the dither that will have the least effect on that sound will be plain old flat vanilla TPDF. That's why many mastering engineers use it more often than the shaped dithers. Shaped dithers sometimes come into play with very quiet, dynamic, or delicate pieces, like a classical concerto perhaps, but even then, some prefer to avoid shaped dithers.