"Bounce-to-Disk" Again...

[Rick Cornish]

I know there have been a number of threads here relating to offline vs. realtime Bounce-to-Disk. The guy who's mastering my album has found a number of instances of **clipping** on my offline BTD tracks which are not present when I go back and do a real-time BTD of the same exact mix.

Ever heard of that?

[HCMarkus]

I always do real-time.

I figure, I've listened to this track probably one million times. It won't kill me to hear it for the 1,000,001th time.

[mikehalloran]

Can't say I've noticed this but I did go to 32bit float for internal processing when I noticed that BTD was giving me distortion on those dense choral tracks I was mixing — big difference with a lot more headroom.

Although my internal processing is normally 48k/32bit float, I bounce to 48k/24 unless it's a CD, then 44.1/16. On the rare occasions I am asked for 96k, it's 96k/32bit float internal and I bounce to 96k/24.

[Rick Cornish]

…and now it turns out I have other problems.

Who knew Waves and Brainworx plugins make such horrible sounds if they go the slightest bit over?

[Tidwells@aol.com]

Forgive my ignorance; Where do I set the internal processing to 32 bit float in DP 9?

Doug

[Tidwells@aol.com]

Now that I'm looking at DP9, I see the little button on the control panel where I can choose between 16 bit integer, 24 bit integer, and 32 bit float. I thought this chose the default bit depth of audio files recorded in the project and that now that DP is a 64-bit application, all internal processing is done with 64 bits and virtually unlimited headroom. Am I misunderstanding it?

Doug

[mikehalloran]

Now that I'm looking at DP9, I see the little button on the control panel where I can choose between 16 bit integer, 24 bit integer, and 32 bit float. I thought this chose the default bit depth of audio files recorded in the project and that now that DP is a 64-bit application, all internal processing is done with 64 bits and virtually unlimited headroom. Am I misunderstanding it?

Doug

Yes, you are misunderstanding it. All Mac apps from Catalina on must be 64bit. This has nothing to do with the internal processing of DP or any other DAW. Some plug-ins process internally at 64bit — again, this doesn't have an effect on DP.

You can set DP to convert all sound bites to your desired sample and bit rate on import as I do. Once you set it in Preferences, it's like old Ron Popeil: "Set it and forget it." You can also open an older project and convert everything in a few keystrokes and a click-drag. Both can be done in Performer Lite, also, but the function is in a different window than DP (why, MOTU?!?!?).

The arguments against using 32bit float boil down to two: 1) file size — for me, the benefit far outweighs this disadvantage and additional storage is cheap. 2) 'no one can hear the difference' — as far as noise floor, this is correct but the headroom improvement can be dramatic. In my case, this proved to be a huge time saver and reduced my need for certain plug-ins. Most of the vocal tracks I was processing were recorded by seniors listening to my guide tracks on headphones and singing into their handheld smartphones.

For three years during lockdown, I can say that discovering 32bit internal processing changed my life. I am now spoiled and will never go back.

Although a handful of DAWs now claim 64bit floating point support, I've not seen the point. Not even Pro Tools or Logic do that, despite the misunderstanding on so many user boards. Their 64 bit summing bus is not the same thing as we discussed in 2021.
viewtopic.php?t=58956&start=15

[Tidwells@aol.com]

Thanks for explaining, Mike! So if I've recorded a song in 24-bit mode, do you think it could still be advantageous to switch to 32-bit float mode before doing the final mix down in order to gain more headroom and avoid any "internal" clipping?

Doug

[mikehalloran]

Thanks for explaining, Mike! So if I've recorded a song in 24-bit mode, do you think it could still be advantageous to switch to 32-bit float mode before doing the final mix down in order to gain more headroom and avoid any "internal" clipping?

Doug

Doug,

Everything I import into DP converts automatically to 48k/32bit float. Doesn’t matter if it’s a CD track, 24bit audio or mp3. The only exception is when the client requests 96k and that’s a simple change in Preferences.

Most of my interfaces are 24bit but I let DP convert automatically when I track—no need to add the extra step. I could use either of my Zoom recorders as 32bit float interfaces, I suppose but I’m fine with just letting them function as recorders and importing the files.

Again, I bounce to whatever rate/depth is required for a project but often, I don’t know what that is so it’s normally 48/24. If I later make a CD, I’m fine with bouncing down to 44.1/16 in DSP-Q. For m4a iTunes, I convert in TwistedWave—though I could keep 24bit, I let it convert to 16bit to save space.

[Tidwells@aol.com]

Forgive my "dense-ness", Mike! I'm having trouble deciding if you are answering "yes" or "no". I already did all the tracking in DP with DP set in 24 bit mode. Would it be advantageous to switch DP to 32 bit float mode before doing the final mix down?

Doug

[CharlzS]

Now that I'm looking at DP9, I see the little button on the control panel where I can choose between 16 bit integer, 24 bit integer, and 32 bit float. I thought this chose the default bit depth of audio files recorded in the project and that now that DP is a 64-bit application, all internal processing is done with 64 bits and virtually unlimited headroom. Am I misunderstanding it?

DP as a 64 bit app is referring to its memory model - what the apps address space is, how much memory the app can access. The other bit reference (24 fixed, 32fp, 64fp) tells you what the math precision is. I believe that DP and most plugs use 32 float internally. Some plugs claim 64 bit math. So any audio recorded in 24 fixed is converted to 32 float automatically for processing, summing, etc. Apparently the only advantage to 24 is that it saves disk space, which is pretty much a non issue these days.

[Tidwells@aol.com]

Thanks for the info, CharlzS! A take your answer as a "no", there will be no advantage in switching DP to 32 bit float before mixing down the audio tracks I already recorded while DP was in 24-bit mode.

Doug

[CharlzS]

Thanks for the info, CharlzS! A take your answer as a "no", there will be no advantage in switching DP to 32 bit float before mixing down the audio tracks I already recorded while DP was in 24-bit mode.

You can bounce out a 32 bit float from a 24 bit file format. I think the difficulty here is that the audio engine is 32 float regardless of what the file format is. When you bounce out at anything less than 32 float, the headroom starts shrinking and additional processing is involved. It's probably OK to leave everything @ 24 bit as long as you know this and bounce accordingly. I've converted some and left some at 24, but anything new is 32 float.

[mikehalloran]

Thanks for the info, CharlzS! A take your answer as a "no", there will be no advantage in switching DP to 32 bit float before mixing down the audio tracks I already recorded while DP was in 24-bit mode.

You can bounce out a 32 bit float from a 24 bit file format. I think the difficulty here is that the audio engine is 32 float regardless of what the file format is. When you bounce out at anything less than 32 float, the headroom starts shrinking and additional processing is involved. It's probably OK to leave everything @ 24 bit as long as you know this and bounce accordingly. I've converted some and left some at 24, but anything new is 32 float.

Everything I work on is 32 bit float. I convert all files in projects older than three years if I'm going to do any work on them at all — takes less than a minute.

Yes, DP process everything at 32bit float but your available headroom is different if your tracks are 16 or 24bit. The more tracks you have, the more complex the waveforms, the more you will notice this. Senior citizen virtual choirs were my baptism on this.

Do you have to convert 24bit files after you have tracked them? Probably not but it depends on the plug-ins you are using and how many of those tracks you have as to whether or not it's a good idea. Since I have DP set to convert everything automatically on import including recording, it's only an issue for me if the project a) is older than three years or b) it originates somewhere else (normally Logic or Performer Lite). Again, I convert everything.

[daniel.sneed]

As Mike said, internal computing within DP is allways 32 bit float, even if you are mixing 16 bit files.

All of DP plugins are 32 bit float. Which is great, for they never clip, whatever level they see.
[Mike pointed to some exceptions. Among them is DP MW Limiter]

Many third party plugins are not 32 bit float. So they can clip and distort whenever peak level exceeds 0dB.

If you are in doubt about this, allow you a very simple test:
Listen to a single DP audio track: if pure and clean sound, then:
Insert 4 DP Trim plugins
First and seconds set to +20dB
Third and fourth set to -20dB
Listen to the track: pure and clean sound

Motu, you rock!