Voxengo Premium Membership - All Voxengo Plugins For a Fixed One-Time Fee
Forums     Plugins     CurveEQ 16 bit vs 24 bit

This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.





I'm rather satisfied with the sound of my 24 bit (44.1) playback.  It sounds as good as some of the better 16 bit pro recordings I've come across.  What bothers me is when I convert down to 16 bit the wav file looses a little bit of the definition and glitter it had in the 24 bit world.  I was wondering if there is anything that could be done in software, maybe in an upcoming version change of CurveEQ, that could compensate for this small but very real loss?  Is it possible to transfer a 24 bit sound quality in to a 16 bit world through some software technique?  I admit this is a small change but I've learned that obtaining very good sounding recordings is the art of maximizing every variable.


Wow, that's kind of the whole enchilada right there, Bob: how to get fewer bits sound like more bits.

I'm afraid it's more than just running your 24 bit files through some version of CurveEQ.  CurveEQ is an equalizer, and that can help in making up for lost sound in those 8 bits.  But it's also about applying dither correctly, gain staging and host of other factors.

I would suggest that you go to Bob Katz's website, www.digido.com, and look up his article about maintaining transparency in your DAW.  That would be a good place to start.  It's a big subject that goes to the heart of good digital sound.  Good luck and join the club!


Bob, you may try some dithering software (not CurveEQ for sure, because it is a different kind of processing).  Pay attention to noise-shaping algorithms they use, because noise-shaping is the only measure which can help to increase resolution in the most important frequency ranges.

I do have a dither capability using CE Pro.  I never use it as I always thought dither was useful only for reducing noise/distortion on very quite passages when converting from 24 to 16 bit.  My stuff never gets dead quite.  The documentation is not clear on the point that dithering is also useful for maintaining a 24 bit quality when converting to 16 bit world.

Thanks for the tips.


Hi Bob,

I use CE Pro sometimes, like now when I'm goofing around with the frequency band splitter to divide a wave into 6 or 7 broadcast bands.

If you're using 24 bit files in CEP then the wave is converted to 32 bit by CEP and when you're finished with the session it may be dithered back to 24 bit depending on your settings.

When you want to master to a medium with a lower bit depth like CD then using dither will help retain some of the info in the extra bits.  Some dither algorithms are better than others.  The Bob Katz book & site are good places for details as well as the Izotope Ozone web site which has a dithering guide.

I can't say that I can notice which dithering is better only that it is pretty obvious if I don't use it, even on loud rockin stuff the 16 bit sound can get kindof gritty or crunchy.  Regardless, you song may fade in at the beginning and fade out at the end - those are spots where dithering will be noticable.  Also any solo reverb tails would benefit, etc.  It just seems like a bunch of dsp calculations made at 32-bit would benefit from some sort of acknowledgement of the extra bits instead of loping them off - that's why I do it.

Truefully kylen, all this stuff about 24 bit recordings edited in 32 bit then saved back as a 24 bit file only to be rendered as a 16 bit file has got me confused - and I've worked with digital electronics for many years.  I experimented with CE Pros dither function and found that the following settings yielded the most 24 bit like sound in my gear at least:

1.  Gaussian (not shaped)

2. bit depth of .5

3.  44.1k optimized

I must admit the idea of less than one bit has me confused.  I know what one or two bits are but a bit less than one?  It is not explained anywhere.  Technically what is less than one bit?  He states a bit depth of .2 to .7 give the best results.

Bob - Important point #1 - I know less about dither than you do I'll bet - I've just read about how to do a couple of things with it and haven't run any comparision tests.  It's an important topic that hasn't bubbled to the top of my list yet :)

I don't know what the .5 bit stuff means but I'll bet the guys over at audiomasters.com would know - it's most of the old crew from the syntrillium forums.  If you haven't been there I'll bet there's either a thread you could search for or one of the more experienced guys would know like SteveG or losts of the other folks.

When I want to try out a real good dither I use Ozone3 dithering.  But I really don't think my AD DA converters and monitoring are at the state now where I could run any conclusive testing with dither - that's why I kinda gotta take everybody else's word for how to do it right now.

Maybe when I get some Mytek or Benchmark grade stuff in here I'll revisit it !

ED: Oops - wrong address the Cool Edit Pro guys are here:


There's some over at homerecording.com too.

0.5 bits means that statistically the noise power added equals to 0.5 bits (this can be called 1 bit peak-to-peak).  Programatically (before rounding to 16 bits happens), it means that added noise is multiplied by 0.5.

Indeed, from my experience, Gaussian dithering offers better results than rectangular or triangular dithering.  However, it is also true that the overall increased noise power is higher for Gaussian dithering than for other types.

Thanks for the explaination Aleksey.  The increase in noise, according to the documention is only 1.2 dB above the triangular model vs. the Gaussian.  The sonic improvement is large enough to overlook this small increase.  All other combinations I tried sounded mostly the same whereas I could hear the improvement with the Gaussian distribution.

I'm looking forward to the next installment of CurveEQ.  I have discovered I get excellant eq-ing results when I plug two instances of CurveEQ into the mains.  I plugin my own eq preset two times, one I leave alone, the other I tweek.  I have found this gives excellant precision and allows me to achieve a higher level of sonic quality than just one instance.  You only briefly passed over this in your help menu but it really should be emphisised as it is a very powerful mastering technique.

Also I would like to point out that I am not happy with the saturation compressor.  It adds two much smearing to my liking.  I would prefer if you disable the smearing/distortion/saturation in the saturation compressor and allow the user to select the level useing the Vintage function if it is desired.


Indeed, that 'saturation' section is pretty smeary/muddy for mastering, but it can still be used for separate tracks.  I had some pretty nice results on the snare, for example.  I have plans for CurveEQ v3 and it won't feature saturation of this kind just to not cause any confusion.
This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.