Jump to content

The Bounce In Place Conspiracy


Arturas

Recommended Posts

After reading this post on dithering - http://productionadvice.co.uk/when-to-dither/ I starting wondering what kind of settings Logic uses for "bounce in place". I recorded a test oscillator sine wave as a note on a virtual instrument (1046.5 Hz, -12 dB). Then I bounced it in place and made versions with all other regular bounce algorithms (including no dither).

I opened them in Izotope RX and looked at the spectral analyser. To my surprise non of them matched the bounce in place file (the screenshot at the bottom). It had the thinnest line with no high or low frequency artefacts.

All the files are 16 bit, but 24 looked the same way. So my question isn't about dithering preferences or the audible difference, they sound the same to my ears. I just would like to know what exactly is happening, I'm getting paranoid! Let's call it scientific interest :wink:

1695047216_BINPLACE.png.fe988db76e102dad9044a601f483f08c.png

2076502451_BNODITHER.png.f2dce9f2ab42b0db5f465fbafddfb8fd.png

657959982_BDITHER.png.8b2ba9ab857bceca0774c37579c6e327.png

1392691358_BDITHERNS2.png.ef7e95567063ee1ca2c00584b3330222.png

373854525_BDITHERNS3.png.54da83abf7293a8d389d21095b329b22.png

1734112828_BDITHERNSUV22HR.png.26c8f30ccd88664858b757d3c090702d.png

Link to comment
Share on other sites

Bounce in Place should be the same as A regular Bounce to 24-bit with the dithering parameter set to none and no normalisation.

That's what I thought too, but the files aren't just different on analyser, the Bounce In Place one is also 9 times larger. How do I share the files here, the forum wouldn't allow WAVs?

Link to comment
Share on other sites

The two files are identical except for the length of the files.

The Bip 1:10 minutes while the other file is 7.5 seconds.

The first 7.5 seconds are identical and null perfectly.

Oh, I'm really sorry for that, I didn't know it does bounce in place until the end of the project, I had cycle on. Anyway, I updated the file and they seem very alike now. But there is still some visual difference. Is it just the RX representation? Or every bounce can be a bit different?

 

Thanks for your help, Eric! And forgive my silliness)

Bounce In Place VS No Dither UPDATED.zip

Link to comment
Share on other sites

It's fine. Your audio quality is fine.

 

Find something else to worry about. :)

I know it's a form of procrastination, but I can't help it)

 

Is there a way to change bounce in place dithering settings though? Other than using a plugin.

Link to comment
Share on other sites

No, you should not use any additional dithering since you are bouncing 24-bit files.

I don't want to be annoying, I'm just trying to get the theory right. Could you, please, explain me why? As I understand, bouncing without dithering introduces some distortion even if I stay in 24 bits (it's not my opinion, I read some Bob Katz articles on this topic and the one I mentioned in the post). Yes, it's very low level, as well as dithering noise, and not audible at 24 bits. But doesn't it build up every time you bounce? What if I do it with 60 tracks in one project?

I also read that Logic does freezing at 32 bit float, which is the recommendation of most well-known mastering engineers (not because of clipping, but the lack of distortion). Maybe it doesn't make any difference sonically, I've seen different views. But why then Logic bothers freezing at this bit depth remaining the only DAW I know without 32 bit float file support?

 

Thank you again for your time. And no offence, I love Logic and your forum has been very helpful!

Link to comment
Share on other sites

I think you have a couple of misconceptions about dithering. Effectively, dithering actually is intentionally applied distortion. It is intended to cover up the "quantize" induced distortion in a summed bounce that occurs as the signal approaches negative infinity (in other words, silence). The distortion involved in dithering is cumulative. That is why you only want to dither once at the very end of the process.

 

I'm very familiar with Bob Katz's attitudes on dithering, he was the mastering engineer on two CDs that I mixed (he's a great guy, and an excellent engineer). His preference (as is every mastering engineer I've worked with) is to receive a mix with no dithering applied so that he can apply the dither at the end of his mastering process.

 

So you definitely would not want to dither tracks that are being bounced in place prior to the final mix. If you are "mastering" the track yourself, you will want to apply dithering once at the very end of the mixing/mastering process applied to the master output. Keep the stem bounces pristine until the final processing.

 

Eric's information is accurate. 24 bit recordings have such a low noise floor, that you won't notice quantize distortion any way. Although it is still considered good practice to dither the final product at any resolution, the real need for dither comes in as you reduce resolution from 24 bit to 16 bit for the standard audio CD master.

Link to comment
Share on other sites

the real need for dither comes in as you reduce resolution from 24 bit to 16 bit for the standard audio CD master.

 

Nitpick mode: The term "resolution" is a horrible one for audio. In this case, you generally dither when reducing word length, so, as GNX says, going from 24bit to 16bit you'd want to dither.

 

Going from 32float (Logic's internal mixer format) to 24fixed is less of a problem, partly because of the dynamic range and partly because a 32f signal contains a 24bit signal and an 8bit scaling factor - the signal is essentially always at 24bit word length, but just scaled up and down according to where it needs to be.

Link to comment
Share on other sites

You are not annoying. Your questions are perfectly valid.

 

Could you, please, explain me why?

 

Because the dithering algorithms in the bounce window are for 16-bits.

There are engineers who are convinced that Logic uses TPDF dither automatically when going from 32-bit floating to 24-bit fixed. I have not seen any confirmation about this but I have not seen any evidence against it either.

The main thing to be aware of here is that the noise floor of quantization noise for a 24-bit file is well below any noise induced by an analog path.

 

But why then Logic bothers freezing at this bit depth remaining the only DAW I know without 32 bit float file support?

 

It uses 32-bit for freeze files to avoid 0 dBFS limit of 24 bit-files (resulting in clipped files).

Why Logic doesn't offer 32-bit floating as an option when bouncing (It does on export) is indeed very peculiar.

Link to comment
Share on other sites

Thanks a lot for your responses, guys, you've made it a lot more clear to me! Now I can drop these crazy concerns and concentrate on music.

 

P.S. I bounced 160 non-dithered sine waves and didn't hear any distortion in the result audio. I might order a t-shirt with this text :D

Link to comment
Share on other sites

  • 6 months later...
The main thing to be aware of here is that the noise floor of quantization noise for a 24-bit file is well below any noise induced by an analog path

 

Actually I don't think this is correct, Bob Katz is very specific on this point. Quantisation distortion caused by moving from 32 bit float to 24 bit fixed does cause distortion in the audible range. Katz is clear that he and many other mastering engineers all agree they can definitely hear it. I have to admit I could never hear it on my mixing speakers (which are very good Adam P11s). But I recently got some Dunlavy mastering speakers (same design as Lipinski) and a Chord amp (Favoured by Abby Rd etc...) and on these, I can hear the difference. When monitoring and A/B 24 bit dithered and undithered I can definitely hear it, you just need good enough monitors and amp. Admittedly it's a small difference, but it's distortion that doesn't need to be there.

 

Kats explains that quantisation distortion tends to collapse the stereo image slightly, lessen the depth of the mix and make things sound "colder and harder" rather than sounding like identifiable distortion. Although here he is referring to 16bit quantisation distortion.

 

If anyone is concerned about bouncing in place not having dither, you can add it using a dither plugin on the channel you're bouncing. Kats recommends PSP X-dither for exactly this purpose. This is also a great plugin to put on your 2 bus to monitor how the dither will sound (as you play back) be it reduction to 24 bit or 16 bit.

 

If Logic does dither on bounce in place, adding a second dither with a plugin will still be way below the noise floor. Kats explains that unlike 16 bit dither, 24 bit dither noise is far below the noise floor (20db below) of even a very quiet DAC. Even 6 consecutive 24 bit dithers are still way below that. Not the case with 32 to 24 bit truncation distortion though, which definitely is in the audible range.

 

I suggest anyone interested in this read Bob Katz book Mastering Audio, he explains all this and much more in great detail. It's a great read.

Link to comment
Share on other sites

  • 1 year later...
You are not annoying. Your questions are perfectly valid.

 

Could you, please, explain me why?

 

Because the dithering algorithms in the bounce window are for 16-bits.

There are engineers who are convinced that Logic uses TPDF dither automatically when going from 32-bit floating to 24-bit fixed. I have not seen any confirmation about this but I have not seen any evidence against it either.

The main thing to be aware of here is that the noise floor of quantization noise for a 24-bit file is well below any noise induced by an analog path.

 

But why then Logic bothers freezing at this bit depth remaining the only DAW I know without 32 bit float file support?

 

It uses 32-bit for freeze files to avoid 0 dBFS limit of 24 bit-files (resulting in clipped files).

Why Logic doesn't offer 32-bit floating as an option when bouncing (It does on export) is indeed very peculiar.

Hi Eric, if I bounce a song to 24-bit and apply one of the dithering options is Logic applying dither to the 24-bit file? Or does it only add dither noise when bouncing down to 16 bit?

Link to comment
Share on other sites

Logic will add dither to either 16 bit or 24 bit.  According to Bob Katz, you should always dither to 24 bit as well 16 bit.  the dither noise for 24 bit is so quiet that it's way below the noise floor of even the quietest DACs.  So there's no harm in doing it.  But Katz makes it clear that you will get truncation noise in the audible range even when bouncing to 24 bit if you don't dither.  Since reading that, I always dither to 24 bit.
Link to comment
Share on other sites

According to Bob Katz, you should always dither to 24 bit as well 16 bit.  

But you shouldn't dither to 24 bit, then take that 24 bit dithered file and dither it again to 16 bits. Here's what Bob Katz himself says: 

 

The Cost of Cumulative Dithering

When feeding processors, DAWs or digital mixers to your recording unit, dither the output of the processor to a 24-bit word. Dithering always sounds better than truncation without dither. But to avoid adding a veil to the sound, avoid cumulative dithering, in other words, multiple generations of any dither. Make sure that redithering to 24- or 16-bit is the one-time, final process in your project.

Source: http://www.digido.com/articles-and-demos12/13-bob-katz/16-dither.html

Link to comment
Share on other sites

David Nahmani wrote:

 

But you shouldn't dither to 24 bit, then take that 24 bit dithered file and dither it again to 16 bits. Here's what Bob Katz himself says: 

 

Strange he does seem to say that in the article you mention.  

 

But in his book Mastering Audio (2015) Katz goes into this subject in great detail, there's a whole chapter on it.  I quote from page 208: "The effect of cumulative dither noise at 24 bits is so low it should not be a concern`: a single dithered digital processor produces -139 dBFS noise, about 20 dB below that of a quiet converter.  Even six 24 bit dithered processors in a row raise the noise floor  to -131, which is more than 11 dB below the noise floor of a quiet converter.  In fact the issue with 24 bit is not the cumulative noise, but whether or not truncation has occurred, which causes distortion (in italics) perceivable above the noise."

 

So... all I can do is shrug on that one :)  

 

My thinking is that when you bounce out a mix at 24 bit to send to mastering, you have to decide whether you want some inaudible dither noise or some audible truncation distortion.  Obviously the mastering engineer is going to have to dither to 16 bit whatever you do.  They may or may not dither for the 24 bit Hi-Res file depending on the mastering engineer and if not you'll have two rounds of truncation distortion at that point (which may well be audible).  My feeling is that a tiny bit of dither noise below the noise floor of any DAC on my 24 bit file seems a small price to pay for avoiding distortion.  Katz also devotes a chunk of this book to explaining exactly what effects subtle truncation distortion has on the sound. He even has test files you can download to hear the difference (which I haven't done).  The difference according to Katz is mainly in the font to back sound stage and stereo image on which truncation distortion has a collapsing effect.  Bob says he can hear even one round of truncation distortion in a 24 bit file.  So I elect to play it safe and always dither :)

 

But yeh, kind of confusing what he wrote in that article.

Link to comment
Share on other sites

Thanks for the replies. The way I understood what Eric was saying is that logic is in fact not adding a low level 24 bit dither while selecting that option at the bounce dialogue. Maybe I just misunderstood.

 

I guess I could do a null test to verify. Looks like it boils down to accumulated dither or truncation distortion.

 

Probably best not to go too far down this rabbit hole of overthinking.

Link to comment
Share on other sites

I like iZotope's dither algorithms a lot, I've done a lot of A/B tests with their different dither types compared to Logic and I always prefer iZotope's.  But I like PSP's even better.  I can't remember if you can do this is Ozone but in PSP's dither plugin or their Xenon limiter (one of the very best out there) you can change the dither type and listen to the difference to see which dither algorithm suits a particular track or album.  Great to be able to A/B them in real time.
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...