John


One thing to keep in mind many plugins process at 64 bits. Sonar PE come with a few but Ozone does and a many others do so as well.

If you can keep the internal data stream at that precision then I would engage the 64 bit audio engine.

You don't need to engage the 64 bit mix engine under Options::Audio::General Tab to get it's benefits.  That option is there so you can "hear" the 64 bit engine when monitoring your mixes, BUT you still get the benefits of it as long as you have the 64 bit engine box ticked in the Export dialog options under "Mix Enables" when you export your mix.  Now, if you bounce internally frequently and you want to keep floating point resolution until the end, before exporting the final mix, then you will want to keep the "Render Bit Depth" found in Options::Global::Audio data at either 32 bit or 64 bit.  I personally chose 32 bit because I find it to be a good balance between high resolution and disk performance, since 64 bit files will tax your HDD more (though that will become a think of the past pretty soon with SSD).  But anyways, that's the way to keep the internal resolution high at all times in Sonar.
 
 
 
@Danny,
 
The guys already answered all your questions, but I just wanted to add that none of this will make a drastic change in the quality of your music.  Things like the performance, the equipment and techniques used to record will always have a greater impact on your sound.  But why not take advantage of it?  It's there, so I use it since it WILL retain the integrety of my music at a much higher level.  And trust me, your machine will handle anything you throw at it.  You're comparing a Pentium D (I believe that's what you had, right?) with a Core i7 - that's not even a fair comparison :-P.  Those two are worlds appart.  In any case, hope you have a great day my friend!
 
 
Take care!
 
 

Jose7822

John


One thing to keep in mind many plugins process at 64 bits. Sonar PE come with a few but Ozone does and a many others do so as well.

If you can keep the internal data stream at that precision then I would engage the 64 bit audio engine.

You don't need to engage the 64 bit mix engine under Options::Audio::General Tab to get it's benefits.  That option is there so you can "hear" the 64 bit engine when monitoring your mixes, BUT you still get the benefits of it as long as you have the 64 bit engine box ticked in the Export dialog options under "Mix Enables" when you export your mix.  Now, if you bounce internally frequently and you want to keep floating point resolution until the end, before exporting the final mix, then you will want to keep the "Render Bit Depth" found in Options::Global::Audio data at either 32 bit or 64 bit.  I personally chose 32 bit because I find it to be a good balance between high resolution and disk performance, since 64 bit files will tax your HDD more (though that will become a think of the past pretty soon with SSD).  But anyways, that's the way to keep the internal resolution high at all times in Sonar.
 
 
 
@Danny,
 
The guys already answered all your questions, but I just wanted to add that none of this will make a drastic change in the quality of your music.  Things like the performance, the equipment and techniques used to record will always have a greater impact on your sound.  But why not take advantage of it?  It's there, so I use it since it WILL retain the integrety of my music at a much higher level.  And trust me, your machine will handle anything you throw at it.  You're comparing a Pentium D (I believe that's what you had, right?) with a Core i7 - that's not even a fair comparison :-P.  Those two are worlds appart.  In any case, hope you have a great day my friend!
 
 
Take care!
 
 

Thank Christ, sombody who gets it's, it's about the music not so much about the sound!!!
I’m not being conceited or arrogant, I just want to give away the formula for turning lead into gold.
Down with the false Gods. (Audio Schools)
Viva la Revolution
Sonar Forever, Pro Tools Never
Ben B.C.T (Bachelor of Creative Technology)
http://www.myspace.com/audiomystics
 

Jose7822

John


One thing to keep in mind many plugins process at 64 bits. Sonar PE come with a few but Ozone does and a many others do so as well.

If you can keep the internal data stream at that precision then I would engage the 64 bit audio engine.

You don't need to engage the 64 bit mix engine under Options::Audio::General Tab to get it's benefits.  That option is there so you can "hear" the 64 bit engine when monitoring your mixes, BUT you still get the benefits of it as long as you have the 64 bit engine box ticked in the Export dialog options under "Mix Enables" when you export your mix.  Now, if you bounce internally frequently and you want to keep floating point resolution until the end, before exporting the final mix, then you will want to keep the "Render Bit Depth" found in Options::Global::Audio data at either 32 bit or 64 bit.  I personally chose 32 bit because I find it to be a good balance between high resolution and disk performance, since 64 bit files will tax your HDD more (though that will become a think of the past pretty soon with SSD).  But anyways, that's the way to keep the internal resolution high at all times in Sonar.
 
 
 
@Danny,
 
The guys already answered all your questions, but I just wanted to add that none of this will make a drastic change in the quality of your music.  Things like the performance, the equipment and techniques used to record will always have a greater impact on your sound.  But why not take advantage of it?  It's there, so I use it since it WILL retain the integrety of my music at a much higher level.  And trust me, your machine will handle anything you throw at it.  You're comparing a Pentium D (I believe that's what you had, right?) with a Core i7 - that's not even a fair comparison :-P.  Those two are worlds appart.  In any case, hope you have a great day my friend!
 
 
Take care!
 
 

Thanks Jose...between you, bit and a few others...I finally get it now. Whew...what cornfusion?! LOL! Thanks again!

now that 32 bits is used for avoiding rounding erors when perform calculation on 24bits, why not has nothing to do with recording bit depth? it inevitablely has something to do with recording bits depth.

I miss reader1's entertaining posts. Almost enough to unblock him. Almost.

I can't believe there are three pages of answers on this simple question.
 
First off, you only dither one time, at the very end.  Keep the dither box in audio options OFF.  The only time you turn it on, is after you have mixed down to a single stereo track, you have "mastered" it and you are preparing it to go on a CD.  That is the ONLY time to use dithering.
 
Secondly, 32 or 64 bit audio engine (processing) has nothing to do with dithering.
 
Third, dithering is such a minute difference in the end product, especialy with the loud, compressed levels we all master with these days.  If you did no dithering at all and just truncated you master to 16 bits, you probably wouldn't hear a thing wrong with it.  Stop concerning yourself with the infinate minutia of "Audio Engineering" and make some music that kicks it!

Butch


I can't believe there are three pages of answers on this simple question.

That's because people keep spreading misinformation. Like:

 
First off, you only dither one time, at the very end.  Keep the dither box in audio options OFF.  The only time you turn it on, is after you have mixed down to a single stereo track, you have "mastered" it and you are preparing it to go on a CD.  That is the ONLY time to use dithering.

  In order to avoid quantization distortion, you should apply dither any time you reduce the bit depth. Somehow this got transformed into "only dither once" which is not entirely accurate. Ideally, you should only reduce your bit depth once, but sometimes you can't avoid it. So if you have to reduce your bit depth, you should apply dither to avoid truncation artifacts.  I don't know of any DSP engineer who would say otherwise.

Sonar is programmed well enough to apply dither when appropriate, and to not apply it when not needed. So you should turn Sonar's dither on, and set it to triangular. Then don't ever worry about it again.  If you want to choose some other dither type, or none at all on mixdown, those choices are available in the export options box.

...
 
Third, dithering is such a minute difference in the end product, especialy with the loud, compressed levels we all master with these days.  If you did no dithering at all and just truncated you master to 16 bits, you probably wouldn't hear a thing wrong with it.  Stop concerning yourself with the infinate minutia of "Audio Engineering" and make some music that kicks it!

I agree with this... as long as no misinformation is being spread. But not everyone creates loud compressed masters.

tarsier

both 24 bit integer and 32 bit float have the same bit depth ... 24 bits ... (23 + sign bit to be accurate) - 32Float adds 8 bits for exponent for range, but same resolution ...

No, 32 bit float has 25 bits worth of resolution. 23 bit mantissa, 1 sign bit, 1 implicit bit. The implicit bit isn't actually included in the 32 bits, that's why it's implicit. But it gives one more bit's worth of resolution.

yes - correct - thanks for pointing this out ....

bitflipper

now that 32 bits is used for avoiding rounding erors when perform calculation on 24bits, why not has nothing to do with recording bit depth? it inevitablely has something to do with recording bits depth.

I miss reader1's entertaining posts. Almost enough to unblock him. Almost.

hehe, you think its very funny?
pls figure out what's wrong with it. you are such an enthusiast. I will correct next time.
just as same as I pointed out that salesdude hasn't cheat you when he speaked out 75W. and what UL is.
 
if you could talk more about dither, I like to discuss either from math or from electronic circuit.
I will never use Artistic discrete converluton formula.
I dont like gossip either.
 
 
 

I'm sorry, but I strongly dissagree. There is no reason to reduce your bit depth at any point in the recording process. What you record at 24 bit, you keep at 24 bit, always. The only time to reduce bit depth, is when it is time to go to a 16 bit CD format. I believe therefor, the missinformation is yours.

When you record at 24 bit, then playback, Sonar upconverts the audio to either 32 or 64 bit floating point format and does its processing. Then you go back to 24 bit for playback. That is an unavoidable bit depth reduction. Dither should be applied just before that point to avoid the quantization artifacts that will occur.  At the 24 bit level, the dither is inaudible as are the truncation artifacts as far as I can tell. But as a matter of best practices, applying dither when bit reducing is one half of a best practice that includes the other half of avoiding all the bit depth reductions that you can. Sonar's 24 bit playback is an unavoidable bit depth reduction. 24 bit export or 24 bit bounce or 24 bit freeze is an avoidable bit depth reduction, but common enough in most workflows.

The misinformation comes in the statement: Only dither once. That statement gets repeated over and over and is a gross oversimplification of the dither topic.  In my book, oversimplification is misinformation and I try to clear that up.

I'll say it again: Sonar is programmed correctly (at least the last time I verified which was around v7 or so...) to apply dither when appropriate and to not apply it when inappropriate, when its dither is turned on. That's why I say turn on Sonar's triangular dither and then forget about it.

bitflipper

now that 32 bits is used for avoiding rounding erors when perform calculation on 24bits, why not has nothing to do with recording bit depth? it inevitablely has something to do with recording bits depth.

I miss reader1's entertaining posts. Almost enough to unblock him. Almost.

I miss bitflipper's hypocritical comments on audio or something like, almost to block him. only a step.
 
I dislike gossip about other. but he seems like it very much.
 

Butch, I'm very happy for you that your workflow is such that you can "only dither once" at the end. I don't have the luxury of that kind of workflow. My projects go through many, many, stages, bounces, transfers, imports, exports, out of my hands, back to my hands, and finally out the door.

Don't call me a liar. I never claimed I could hear one stage of 24 bit dither vs. no dither. However on one of my projects that went through numerous of those stages I outlined above, I started to notice a graininess in the reverb tails like light cellophane in the breeze. How could this be? I thought. I checked the dither settings (among lots of other things) and they had been turned off throughout the project. So my numerous processes had been getting truncated to 24 bit, each time adding a bit more grunge instead of just smooth noise. Eventually the grunge was audible.  So I rolled back to the original sessions, re-did everything and the tails smoothed right out.

And when you say "if you keep your dither on, and it's doing all of this important inaudible work at the "bottom" of the 24 bits, as soon as you go to CD, you are going to lop off all of that minute, undescernable dithering all the way up to the 16th bit, making all previous dithering a waste. "  That's just not true, and a complete misunderstanding of what dither does.  Frankly, if you believe that, then you wouldn't even think dither is necessary at all when you go to 16 bit.

The CPU hit of triangular dither is so close to 0 as to be 0. Irrelevant argument.

Again: "only dither once, at the end" is a gross oversimplification at best and misleading at worst. Feel free to believe and do what you want. But I'll also feel free to correct your misinformation. Set Sonar's dither to triangular and don't worry about it.

Part of the complexity of this is there's 2 pieces to it:

1. What's the best way to theoretically preserve as much of the low level signal as possible?

2. Is what you're preserving at such a low level that it's even audible?

I believe much of the arguing is about #1. 

If, in #2, it turns out to be inaudible - as in some (many? most?) cases it would - it's really a bunch of arguing about "nothing".