Tracking at 32 bit float. Clipping?

[cron]

Slightly daft beginners question here that I haven't been able to solve and don't have the necessary equipment at hand to test.

A friend of mine makes radio using a modest (by broadcast standards) set-up. Broadcasts often up to 3 hours long. A lot of these are pre-recorded live (i.e. no multiple takes). As such, he finds staying out of the red while both minimising his (considerable) noise floor and maintaining an average volume a constant battle when juggling vinyl/CD/digital over extended periods of time. His digital audio knowledge is minimal.

I was wondering if tracking at 32 bit float might help him avoid minor clipping artifacts in these circumstances. Obviously his converters won't utilise anywhere near this kind of resolution, and I shan't be advising him to crank the volume and forget about it, but I was wondering if this might be able to save him from the odd clip here and there. Changing the volume of a track halfway through doesn't sound great on radio!

I tried Googling this but just got loads of threads along the lines of "is 24 bit better than 32 bit" etc. I only save to 32 bit float when using offline renderers like Hourglass/CDP etc, and I've no idea if those magical clip-avoidance properties I find so useful inside the box translate to recording in from outside the box. Is full scale ITB also the full scale of your A/D converter, or is there any room to play (however small) above this?

Any takers?

[Keith99]

Any digital clipping, whatever the bit depth used, is to be avoided. I am surprised he has such a bad noise floor - what is this caused by?

[No_Use]

I was wondering if tracking at 32 bit float might help him avoid minor clipping artifacts in these circumstances.

No it won't.
It's the resolution of the converters of the soundcard that matter when doing analog-digital conversion during recording and they are 24bit (or 16bit on cheap/built in cards).
When you 'track' in 32 bit float you still get 24 bit resolution from your soundcard (meaning > 0db FS = clipped), the file just gets filled with zeros in the DAW while being converted to 32bit float files. So it would additionally be a waste of HD space to record in 32bit float.

(Though I've heard it might be easier on the CPU if processing these files afterwards with FX in the DAW, because the internal conversion from fixed to float which (almost ?) all DAW's do nowadays doesn't need to happen anymore, but not sure if that's correct).

(Different thing, as you already said, if you render from within your DAW to a float file, than you can go over 0 db FS without clipping because no A/D or D/A conversion happening there.)

[Panphobia]

On a similar note, would it be useless to use 32 float ITB with a 24 bit card? The idea being that calculations are done in 32, even if I don't hear them coming out of the speakers. Or is everything just truncated anyway?

[cron]

I am surprised he has such a bad noise floor - what is this caused by?

A ragtag mish-mash of consumer grade DJ equipment.

No it won't. It's the resolution of the converters of the soundcard that matter when doing analog-digital conversion during recording and they are 24bit (or 16bit on cheap/built in cards).

Many thanks. I thought about it a bit after posting and realised that, for my bonkers notion to work, the DAW itself would need to be doing the A/D conversion which is of course impossible. Many thanks.

[cron]

On a similar note, would it be useless to use 32 float ITB with a 24 bit card? The idea being that calculations are done in 32, even if I don't hear them coming out of the speakers. Or is everything just truncated anyway?

Most DAWs will be using doing floating point calculations behind the scenes regardless of the audio you supply it with AFAIK. You could be working with 8 bit audio and it'll still process it as float. Your 24 bit card has no effect on your DAWs internal calculations.

[ZenPunkHippy]

I guess you probably know this already, but to achieve consistent loudness for radio broadcasts the stations use compression / limiting. That's part of the reason we have the loudness wars ...

So your friend really needs to combine the music / ads + voice in to a single stream with a compressor limiter on it.

I would guess that radio stations might also have someone dedicated to riding the fader of the presenters voice, or at least on hand to adjust for loud guests.

Peace,
Andy.

[cron]

I guess you probably know this already, but to achieve consistent loudness for radio broadcasts the stations use compression / limiting. That's part of the reason we have the loudness wars ...

So your friend really needs to combine the music / ads + voice in to a single stream with a compressor limiter on it.

I would guess that radio stations might also have someone dedicated to riding the fader of the presenters voice, or at least on hand to adjust for loud guests.

Peace,
Andy.

Aye, one of the stations he sends stuff over to compresses everything. Drives him nuts! He has got a 'fader rider' over at one of the places he does live shows though. Friend does the guy who is on before him, the guy who is on before him does the friend.

I've tried to convince him to invest in better gear, but deaf ears. Like with every addict... sweet, sweet virgin vinyl always comes first.

[ZenPunkHippy]

Thanks for the "quoted out of context" quote

I think what he should be doing is tracking through a compressor or limiter i.e. destructively. Doing this he can keep the input volume on the sound card at a level that avoids any clipping, but pump the gain using the compressor / limiter. Most hosts (Reaper if needs a cheap solution) can route audio in such a way that the FX are applied to what's being recorded.

Is that too obvious? Plenty of free options for comp/limiting if he wants to continue investing in vinyl

Peace,
Andy.

[cron]

Got some compression on his voice going in, but that's about as destructive as he's willing to go.

That quoted out of context thread is brilliant ! Completely passed me by. Need to look in the Off Topic forum once in a while.