Difference between 24 bit and 32 bit recording.

[jmbriley]

What is the difference between 24 bit and 32 bit audio recording in Tracktion as far as quality if you are using a 24 bit sound card. Thanks.

[msl]

I am surely no expert but I can't see any advantage to recording a 24-bit source in 32-bit. It's like counting the coins in your pocket to the nearest 100th of a cent/penny/whatever-your-smallest-coin-is.

Might be good for stress-testing your processor, memory and hard-disk with all that extra data though.

[IIRs]

AFAIK recording 32-bit files will slightly reduce the load on your cpu (as there is no need to convert to 32-bit float for the mix) at the expense of increased disk activity and (possibly) larger files.

As long as you avoid 16-bit you'll be fine!

[jtxx000]

As long as you avoid 16-bit you'll be fine!

i record at 16 bit right now... what's the advantage of 24 bit besides the fact that it sounds better?

[IIRs]

what's the advantage of 24 bit besides the fact that it sounds better?

That's a good enough reason for me!

24-bit provides much better quality with low level signals, so removes the need to drive the converters with hot signals to maintain resolution: when my VS880 was the hub of my studio I found I could get noticably smoother sounding results if I ignored the clip LED (which lit about 3dB down as a warning) and set the levels according to the meters on the LCD screen.. but it meant either risking clipping during a great take that went a bit louder than expected, or protecting the ADC with a limiter (which could sometimes do more damage than a short clip!)

Recording 24 or 32-bit signals means you can leave as much as 12dB of headroom while still recording ~22 bits of meaningful information (as 1-bit roughly equates to 6dB of dynamic range).. it still just comes down to sounding better though!

[braj]

platinumears,

are you saying that recording at a higher bit rate will put LESS stress on the CPU and more on the HD? I have a older laptop that has a fast 7200 8 MB HD, would I do better at a higher bitrate ie. record more tracks before needing to freeze? The HD currently doesn't get taxed too much. That's what it sounds like you're saying, but somehow that doesn't make sense to me. Can you clarify?

[jtxx000]

platinumears,

are you saying that recording at a higher bit rate will put LESS stress on the CPU and more on the HD? I have a older laptop that has a fast 7200 8 MB HD, would I do better at a higher bitrate ie. record more tracks before needing to freeze? The HD currently doesn't get taxed too much. That's what it sounds like you're saying, but somehow that doesn't make sense to me. Can you clarify?

if i understand correctly tracktion uses 32 bit floats while 16 bit wav files are ints. tracktion has to convert the ints to floats in real time to play back... still i can't imagine it being that taxing on the cpu.

[IIRs]

are you saying that recording at a higher bit rate will put LESS stress on the CPU and more on the HD?

I would never say that, as I would more correctly say bit-depth.

I'm only talking about going from 24-bit to 32-bit as the 32-bit files are already in the correct format for Tracktion's engine. I can't imagine it being significant, but it wouldn't be too hard to set up a test.. let us know your results if you do!

[braj]

I think the added file size would be a deterrent for me, and I'm currently just using 16 bit anyway. I just wanted to understand the mumbo-jumbo

[AndrewSimon]

Difference between 24 bit and 32 bit recording?

8 more bites

[rich_h]

Difference between 24 bit and 32 bit recording?

8 more bites

Extraneous 'e' alert!

[C00kie]

I just wanted to understand the mumbo-jumbo

Most CPUs nowadays work most efficient at 32bits. The whole architecture is set up for 32bits.

If you have a 24bits data stream, then possibly a lot of alignment has to take place. Here are four 24bits samples, three bytes big each, before alignment:

Code: Select all

aaa bbb ccc ddd

These are stored to and fetched from the harddisk in blocks of 4 bytes:

Code: Select all

aaab bbcc cddd

But the blocks need to be realigned, so each sample is separated:

Code: Select all

aaa0 bbb0 ccc0 ddd0

And when stored this process is reversed. It takes some CPU, but not enough to really worry about.

I hope this makes some sense...

[jens]

anyone who thinks dithering is of any importance should
at least record at 24bits (if he's got the soundcard for it)

[Sage]

platinumears,

are you saying that recording at a higher bit rate will put LESS stress on the CPU and more on the HD? I have a older laptop that has a fast 7200 8 MB HD, would I do better at a higher bitrate ie. record more tracks before needing to freeze? The HD currently doesn't get taxed too much. That's what it sounds like you're saying, but somehow that doesn't make sense to me. Can you clarify?

It's pretty simple. Tracktion does all of its mixing internally at 32 bits. That means that every track that is not already stored in 32-bit resolution has to be converted from whatever format it's in. That takes up more CPU than just playing back a 32-bit file. How much more, I wouldn't be able to tell you.

The downside is that 32-bit files take up more space on your hard drive than 24-bit files, and require more disk bandwidth to play back. So you're trading disk bandwidth for CPU performance.

[Sage]

anyone who thinks dithering is of any importance should
at least record at 24bits (if he's got the soundcard for it)

There's a thought -- if I record a track at 24-bits, but Tracktion is processing it at 32-bit precision, is Tracktion dithering the file down to 24 bits, and then converting it back to 32 during playback and mixing? That is to say, would a 32-bit file be a more accurate representation of the signal that was recorded?