Using 32bit Float Point

[DonaldM]

Say I have been sent an audio file that clips due to recording too hot. Assume also my session is set up as 24bit, 48k sample rate. Can I export the clipping clip as 32bit and then import back into my 24bit session and be able to normalize it back from the dead? Or does my session have to be 32bit to start with to do this?

[WombatStudio.Org]

You cannot take a file that is clipping and magically make it not clip.

Pushing into 32 bit float as clipping you might be able to normalize if that's the source of the clipping, but if it's a recorded clipping source ... again, no magic.

[DonaldM]

Originally Posted by WombatStudio.Org

You cannot take a file that is clipping and magically make it not clip.

Pushing into 32 bit float as clipping you might be able to normalize if that's the source of the clipping, but if it's a recorded clipping source ... again, no magic.

I kinda suspected that, but that's why I asked. In this case, I know it wasn't recorded 32bit. Thanks for the quick response guys!

[originalscottyg]

Even if it was recorded in as 32bit float, if it clipped on the way in, they it isn't recoverable by simply reducing the levels, as the converter hardware was clipped during the A/D conversion. The only way this really works is if the audio is clipped in software in a 32bit float session, then output as a 32bit float file.

[feliponk]

Originally Posted by kdarbyshirebryant

Only floating point files can code values above 0dBFS (beyond clipping)

I think you're wrong, in digital there's nothing over 0 dBFS. 0 dBFS is assigned to the maximum possible digital level.

http://www.soundonsound.com/sos/jan0...s/qa0108_3.htm

[Darryl Ramm]

32 bit floating point absolutely allows for greater that 0db digital signals internally to the DAW, of course none of this internal processing affects what a fixed point/fixed dynamic range DAC can output, but you can easily see this in operation by mixing/processing a hot signal at >0dB and then pulling the level down on the master fader to avoid clipping the output DAC. Whoa and the signal is recovered... Which would not be possible if internally the 32 but float could not represent greater than 0dB. This is the whole point of wanting that 32 bit float format.

[DonaldM]

Okay, so now I'm confused. Say I have a clipped file that was sent to me 24bit. Can I import into a 32bit float session, convert it, and then recover it? Is that what you're saying Darrel?

[Darryl Ramm]

No--as others here have said. You cannot invent information/data out of thin air. If fixed point data is clipped, it is clipped, data gone/not there. You cannot magically bring back missing data. But 32 bit float formats help avoid you losing data when things would otherwise clip during digital processing inside Pro Tools (or just about every other DAW).

[feliponk]

Originally Posted by kdarbyshirebryant

I think you need to read the article you quoted carefully. It discusses quite clearly the ability of floating point to code values above 0dBFS. The problem comes at output or transport where values are restricted to integer 16 or 24 bit maths.

I read it well, and still 0 dBFS is the max output on the digital domain.

[Darryl Ramm]

Originally Posted by feliponk

I read it well, and still 0 dBFS is the max output on the digital domain.

Fighting semantics here, but that is really in the analog domain (or when truncating to fixed point). Outputting in the "digital domain" with 32 bit float would be sending somebody a 32 bit float file. And that can contain values that represent more than 0dbFS.