PlanetZ forums - scopeusers.com

32 bit recording of vocals is amazing - I tested it once - recorded 24 bit, 32 bit w/ the 6, 12, & 18 saturation...and there is total difference!
18 is too much - sounds like clipping almost but the 12 is perfect...it really warms up the sounds and makes all nice and clean...

that's true, we also noticed a difference between 24 & 32 bits in vdat... more headroom i was literally srceaming at the microphone and the other end held it up; no clipping! then we also noticed that logic didn't support 32 bit recordings this product is so cool

<font size=-1>[ This Message was edited by: legros on 2003-09-08 11:33 ]</font>

How on earth do you use VDAT for anything? It's editing capabilities are not too great... and it's not that easy to use (compared to logic or Cubase). Or I'll have to read the manual

well i don't

Hello to all people!

If you want to record in 32bit (not floating-point) and dont want to spend extra money on VDAT, you could use the STS samplers...
Furthermore, SoundForge does 32 bit too.. and you can sync it to Cubase or Logic or whatever so you can then easily import the audio tracks into the sequencer and everything be in sync. (that is in theory...I have NOT done it in practice)
The only thing I am not sure, is if the 32bit recording in SoundForge, is at true 32bit resolution since the WAV modules of SFP can either be 16bit or 24bit... Does anybody know?

It is true that 32 bit float is the format widely used by most software, though the resolution should be the same wether it be float or integer. Why Creamware choose integer and not float I do not know, as I have heard the Sharc model they use can do both, but maybe not so well.

Anyway. SFP is 32 bit all the way through.

As I understand it, 32 floating point is actually only 24 bit sound - but then it is 24 bit sound all the time. If you lower the signal 12dB, you don't loose 2 bits - you just get you 24 bits at a lower level.

That's a good way to descibe it
That's the reason why SFP, SonicHD and ProTools|HD sounds better than the software mixers:
32/48bit fixed versus 24bit (=32bit float)

Kim.

interesting.
so the advantage of 24bit-float isn't just important in case of recording but also after it when just lowering the level in the mix, is that right?
(I'm a nerd in this)

Adding to this, how bout the difference between 44.1 and 96Hz. I haven't investigated 96Hz properly yet, so has anyone any views?
Great thread people.

hubird

Floating point has one of its great strengths, when you want to lower the volume os something. If you lower a 32 bit flat signal by 48dB, it will be 24 bit flat. If you lower 32 bit floating by 48dB, it will still be 24 actual bits. So in most situations, you will have the most bits if you work with 32 bit flat compared to 32 float. Only when you get below -48dB, will you get more bits out of floating point.

One nice thing, when working with floating point (Samplitude) is, that if you edit a wave file, so it exceeds 0db, it will not be flat topped - you can just normalise it down, and it will be fine.



Richardo

IMO a new thread would be nice for that.

Done!

Yes, I think they call the 8 extra bits in 32-bit float for "scaling bits".

Am I right in saying that you never lose any information (be it good or bad) using 32-bit float while you can cut away the lower dB's using 32-bit integer during mixing? Thus recording a hot signal in 32-bit integer and lowering it will remove some of the background noise?

If you cut away so many bits, that you start to get rid of your background noise, you are very likely cutting away so much signal, that the quality starts to degrade.

Guys,

A 32-bit file does have the recorded 24-bits plus 8 exponential bits.. Whith float calculation you can generally get very large and very small numbers. You might get a lower noisefloor and processing that was done whith a 32-bit file does not truncate that easily... Whith 32-bit the sound holds up a bit better after repeated processing, and technically that also makes sense. If you start with a 24-bit file and do a process to it, Cubase does the processing at 32-bit then reduces that back to 24-bits on disk...

The output of most mixers is also 32-bit. "Overs" on the master fader won't distort the file and it's a better thing to start from if you want to process the mix files after the fact. But at some point it has to get rounded or truncated down to at least 24-bit, and everything you hear in the speakers will reflect that process, even if you're listening to a 32-bit file. So the benefits are not obvious at first sight.

Nothing I´ve heard so far has a true 32-bit dynamic range... Like garyb said, you need a perfect room, mic, pre, converter, isolated cables, etc. Therefore some engineers prefer 16-bit as format, especially when recording noisy things like drums etc. For straight recording, I can't tell a difference between a 32-bit file and 24-bit, and technically there shouldn't be one. But especially when doing lots of offline processing, 32-bit can be the preferred resolution...