Am I right to think that audio editors with high bit depth processing don’t exist yet? At least that’s what I’ve always thought. Sound Forge 7.0 currently works with 24 bit processing and I’m wondering if there might be some other software out there already that have some technical improvements over Sound Forge.
I guess the main problem is really that computers might not really be up to the task yet of handling 32 or 64 bit floating point precision in real time, but it would be cool if it was available at least as an option so that you could render effects such as compression on EQ on files without having to worry about dithering right until the end when limiting your tracks. Maybe I’m just dreaming.
The thing is that audio editors claim they support bit depths as high as 32 bit float or 64 bit float in their specifications (Sound Forge does anyway), but it actual fact they’re talking about the supported bit depth for creating new samples. But I’m talking about the actual processing engine. Like when you apply a DX effect to a wave file and the dialogue window says “(24-bit processing)”.
When applying, say, compression or a limiting to a wave file that has a bit depth of 32 bit float, you still have to enable dithering on the plugin because the host passes the audio along in 24 bit. The same goes for Sony Vegas 5.0, and I just want things to be passed along in 32-bit float so that dithering isn’t needed.
Waves’ LinEq.pdf states the following:
As the LinEQ process is a double precision 48bit process, the output is rounded back to 24bits. While the equalization doesn’t produce quantization error and noise, the rounding back to 24th bit may. It is On by default, but it is the choice of the engineer whether to add low level hiss like noise or to get slight low leve, program related, non-linear distortion from quantization error. Either noise types will be extremely low and rather inaudible.
There are some native Hosts applications where you know the plug-ins output is passed on in higher then 24bit, i.e. 32bit FP. In these cases it is actually recommended to turn the Dither off and use dithering only at the very last process of the chain directly to the destinatinon bit resolution using L2.
That’s where I’m getting at. They mention that these hosts exist, but as of yet I’ve never come across them.
I think its actually the other way around…
This is direct quote from the WaveLab 5 website…
''Uncompromised audio quality with up to 32-bit and 192 kHz resolution and Apogee UV22HR dithering ‘’
''Support of a multitude of different audio file formats (WAV, AIFF, AU, MP3, MP2 (M.U.S.I.C.A.M.), RAW, Windows Media 9, etc.) and resolutions (8/16/20/24 bits with up to 192 kHz) ‘’
So in this case, it supports formats of up to 24bits, but does calculations at 32bits.
24 bit is an integer format, the 32 bit floating point format (usually)
uses 24 bit for resolution and 8 bit for order of magnitude.
I guess this is why 24 bit is enough for storing, as long as the samples
are kept within the range ± 1.0 those 8 bits extra doesn’t give more detail?
Correct me if I’m wrong, but that’s my belief.
