16 vs. 24 vs, 32(floating) bit recording

[Walter Tore]

I have a delta 1010 and can record at these rates. Could anyone give a basic breakdown of how the sound quality of each rate differs, and hard drive use differs with each? Thanks! Walter

 

[dgatwood]

I have a delta 1010 and can record at these rates. Could anyone give a basic breakdown of how the sound quality of each rate differs, and hard drive use differs with each? Thanks! Walter

Sound quality.... 16 bits if levels are really exact is CD quality. 24 bits lets you ride in the mud and still get close to CD quality. 32 bit floating point is mathematically capable of holding... roughly 24 bit fixed point data without significant precision loss, IIRC, but has the advantage of being practically immune to distortion from saturating a bus when mixing (albeit with increasing precision loss as you get farther beyond the realm of sanity, IIRC).

Almost all software that does effects processing uses 32-bit floating point for the effects path for this reason. Very few programs store data on disk in 32-bit floating point format because as a disk format, it doesnt' really offer any real benefit over 24-bit integer.

As for the storage requirements:

16 bits: (2 bytes/sample) * (n samples/second) = 2n bytes per second
24 bits: (3 bytes/sample) * (n samples/second) = 3n bytes per second
32 bits: (4 bytes/sample) * (n samples/second) = 4n bytes per second

So if you have 48 kHz, you're usign 96kBytes/sec.@16 bits, 192 kBytes/sec.@32 bits.

 

[Walter Tore]

dgatwood: Thank you very much for that response. Walter

 

[Farview]

The difference between 16 bit and 24 bit is pretty big. Most DAWs do all the processing at 32 bit, storing at 32 bit means that the DAw doesn't have to convert on the fly. However, your hard drive has to work harder. I just store at 24 bit.

 

[dgatwood]

The difference between 16 bit and 24 bit is pretty big. Most DAWs do all the processing at 32 bit, storing at 32 bit means that the DAw doesn't have to convert on the fly. However, your hard drive has to work harder. I just store at 24 bit.

It's not necessarily that hard to do the conversion.... A fairly easy conversion routine looks like this:

http://developer.apple.com/document....html#//apple_ref/doc/uid/TP30000732-BAJCBIAF

Bonus points if anybody finds the typo in the second example before the corrected version goes live in a few days.

 

[amoeba]

typo in the second example ...

There's an 8 where there ought to be a 7.

(Back on topic) Walter, you really want to use the 24-bit mode of your 1010 if at all possible, rather than the 16-bit mode. This allows you to record at more conservative levels without sacrificing quality. You get shedloads of extra headroom and don't need to worry so much about clipping. Then you can control the dynamics at the mixing stage instead of at the tracking stage.

a.

 

[altiris]

Since your on this subject; I've never heard anyone mention 32bit 96k sample rate. Is there a reason for that? I mostly here 24 96 but not 32 96.

 

[dgatwood]

Since your on this subject; I've never heard anyone mention 32bit 96k sample rate. Is there a reason for that? I mostly here 24 96 but not 32 96.

It basically isn't a real sample rate. I don't know of any hardware that natively uses floating point audio data, so if you're recording in 32-bit float format, you're probably really recording 24-bit integer, with 32-bit float conversion being done at the driver level or audio framework level or whatever. It's basically just a pretty little lie to make us feel better about ourselves.

 

[altiris]

well like in cubase the default is 32 float 44.1k. so I can leave the 32 as is and change the sample to say 96? and that wont be any differa=ent than 24 96?

 

[Chris Shaeffer]

Said a different way...

You have 24 bit convertors. Period.

So you can't actually get higher bit depth than 24 bits. The other 8 bits are used for processing within the DAW and really just takes up extra space on disk if you want to store it as 32 bit float. So 24/96 is the same as 32/96 for all practical purposes.

There are some input effects in Cubase SX that require 32 bit float... but why add effects at input when you have so much control and so many options if you wait?

Take care,
Chris

 

[Walter Tore]

There's an 8 where there ought to be a 7.

(Back on topic) Walter, you really want to use the 24-bit mode of your 1010 if at all possible, rather than the 16-bit mode. This allows you to record at more conservative levels without sacrificing quality. You get shedloads of extra headroom and don't need to worry so much about clipping. Then you can control the dynamics at the mixing stage instead of at the tracking stage.

a.

Will that mean better sound quality in the finished product? I recorded last night at 24 bit, and it sounded deeper than 16 bit, but to burn it to cd, via my nero program, I had to convert the wave file to 16 bit. Thanks everyone! I am learning. Walter

 

[Walter Tore]

Will that mean better sound quality in the finished product? I recorded last night at 24 bit, and it sounded deeper than 16 bit recordings I have been doing for the past 2 years, but to burn it to cd, via my nero program, I had to convert the wave file to 16 bit. Can you burn a cd at 24 bit? Thanks everyone! I am learning. Walter

Can anyone answer this one? Thanks. Walter

 

[Chris Shaeffer]

You can't burn to a regular audio cd at 24 bits. CD's are 16/44.1, period.

The higher the quality of the tracks that your feed the finaly 16/44.1 mix, though, the better it will sound.

-Chris

 

[Walter Tore]

You can't burn to a regular audio cd at 24 bits. CD's are 16/44.1, period.

The higher the quality of the tracks that your feed the finaly 16/44.1 mix, though, the better it will sound.

-Chris

thanks Chris. That is what my ears were telling me. Walter

 

[Chris Shaeffer]

Walter, if you don't know what dithering is already look into it. Its the highest quality way to reduce bit depth. You may want to figure out if your current conversion process includes dither and figure out how to add it if it doesn't. Chopping a 24 bit mix down to 16 bits doesn't work all that well.

-Chris

 

[Walter Tore]

Walter, if you don't know what dithering is already look into it. Its the highest quality way to reduce bit depth. You may want to figure out if your current conversion process includes dither and figure out how to add it if it doesn't. Chopping a 24 bit mix down to 16 bits doesn't work all that well.

-Chris

Hi Chris: I have never heard of dithering. I will look in my samplitude manual, to see if it is in the program. Thanks again! Walter

 

[Walter Tore]

Hi Chris: Thanks for that info. Samplitude automatically does this, lucky for me! It allows you to change the settings, but the 24-16 conversions sound good to me. Here are 2 songs I did 24-16 bits. All other songs on the site are recorded in 16 bit. I am getting there, thanks to kind people like you, here on the forum. Thanks! Walter
http://www.soundclick.com/bands/pageartist.cfm?bandID=157137

 

[altiris]

You can't burn to a regular audio cd at 24 bits. CD's are 16/44.1, period.

The higher the quality of the tracks that your feed the finaly 16/44.1 mix, though, the better it will sound.

-Chris

So what about DVD? my DVD player has 24/96 on the DVD tray. If you burn a CD or DVD at that (thats if the software permits you) will the DVD player play it? I cant try it for testing since my DVD player is old and cant read disks I burn. Thats another thing. If it says 24/96 I asume DVD movies audio is at that rate cause it just sounds so much better than an audio cd which is 16/44.

 

[Chris Shaeffer]

Well, I'm no expert on DVD audio but there's a couple reasons DVD's sound better than CD's.

1) Dynamic Range: Movie soundtracks actually use it! They aren't squashed into the top 6dB like most commercial CD's. This is probably most of what makes it sound better, in addition to the...

2) Higher Resolution: As you say DVD's are higher res. and that lets the lower volume stuff maintain its integrity. Just going to 24 bit is most of the difference that's easily heard- and the jury is still out about how much benefit there is to really high sample rates, especially with pro-sumer gear.

Most CD/DVD burning software I've seen recently has some sort of option to burn DVD music disc, but I'm not sure how well it works. I just tried it with Toast on my computer and it handled 24/96 files just fine.

Anyway, I wouldn't go overboard worrying about 24/983472374562398567... unless you're working video and are going to end up at a higher samplerate.

-Chris

 

[BigRay]

file sizes

PCM .wav file sizes
MB / Hour =
bits / sample samples / sec bits / byte bytes / KB KB / MB seconds / hour # of channels MB / hour MB / sec GB / hour
16 44100 8 1024 1024 3600 2 605.6 0.168 0.591
16 48000 8 1024 1024 3600 2 659.2 0.183 0.644
24 48000 8 1024 1024 3600 2 988.8 0.275 0.966
24 96000 8 1024 1024 3600 2 1977.5 0.549 1.931
24 192000 8 1024 1024 3600 2 3955.1 1.099 3.862
Short-hand:
24 bit, 48 kHz ≈ 1 GB per hour
24 bit, 96 kHz ≈ 2 GB per hour
24 bit, 192 kHz ≈ 4 GB per hour
sample second 8 bits 1024 bytes 1024 KB min hour