TranceAddict Forums (www.tranceaddict.com/forums)
- Production Studio
-- The 0db limit?

The 0db limit?

It's something we all just accept, that we can't go over 0db with digital audio.

But why? Was it a standard set by some organisation? Is it a hardware/electronic limitation and if so wouldn't that mean it varies from system to system?

I don't get it! Anyone know the reason behind the 0db limit?

It's a mystery to me.

dB is a reference scale. It refers to a gain. Every reference scale has to be in reference to something, and in digital audio the reference point is the maximum possible amplitude (i.e. all 16 bits set to 1). 0 dB means that there is no gain - the signal is at exactly this level. Anything lower than 0 dB means that the gain is negative and that the signal is therefore less loud.

The actual volume associated with 0 dB does vary from system to system, again because 0 dB refers to the maximum volume it's capable of. There's another reference scale for physical sound waves and that's dBA (not actually the correct symbol but a common shorthand), which uses a weighting based on the human ear. You also see symbols like dBV or dBW (in reference to 1 V or 1 W respectively).

So the reason? Well, what else would you use as a reference point that would actually make sense? There is no physically measurable unit here like a volt or watt, so there aren't really any other options.

Re: The 0db limit?

quote:

Originally posted by Synchronicity It's something we all just accept, that we can't go over 0db with digital audio. But why? Was it a standard set by some organisation? Is it a hardware/electronic limitation and if so wouldn't that mean it varies from system to system? I don't get it! Anyone know the reason behind the 0db limit?

This is simple.

Ask yourself.... what is digital audio. It is a series of samples (words) that are all played concurrently. 44,100 per second.

So... your word length is is 16-bit (or 24, whatever). each word has 16 bits to fill up. they're either 0, or 1.

When all the bits are 1, you've reached 0dbFS. When you try to cram to much into a signal, you fill up to 0dbFS, and the remainder is digital distorion that sounds completely horrible. thus, stay away from 0dbFS.

in analog gear, its a different story. you can overload a tube, but it distorts, its not a black or white set value like the digital signal.

good read:
http://www.amazon.com/Mastering-Aud...ie=UTF8&s=books

quote:

Originally posted by camsr It's a mystery to me.

Well I just found a page that seems to explain it, but I'll need to go over it a couple of times.

Here's a quote:

"Sometimes you might wonder what is the dynamic range of a digital audio system. Hard to use accurate definition is that the total amount of uniquely representable information in thesystem is, indeed, proportional to the dynamic range inamplitude multiplied by the bandwidth. But, if you accept the definition of dynamic range as " theratio between the largest possible undistorted signal and thesmallest unambiguous change in signal," then it's easy.Actually, for run-of-the-mill PCM data, dynamic range only depends on the number of bits.To do the bare calculation, the dynamic range in dBs is20 log (2^n) where n is the number of bits. But as I say, that is farfrom the whole story. The generally accepted rule of thumb is 6 dB per bit.So a 10-bit system has 60 dB, a 16-bit system 96 dB, etc. Please note that this value is for a single sample. Through common averaging techniques, it is possible, just like in non-digital systems, to encoded and detect signals that are well below the noice floor. The 96 dB figure is the worst-case, non-ditheredinstantaneous dynamic range of a 16-bit system. When talking about A/D-conversion, things get more complicated. 0dB is easy - it is the biggest sine wave you can fit into the digitalspace. You are driving the ADC end-to-end. The other end is more problematic. You find it by measuring the noiselevel. Unfortunately this depends on the measuring bandwidth.Every time you halve the measuring bandwidth, you reduce the noise by 3dB. Of course you can simply include the entire bandwidth, but in the case of audiothat isn't really fair, because we don't hear that way. Particularly at low levels, we hear predominantly the middle range frequencies, andthe extreme lows and highs disappear. At threshold levels even "A"weighting gives unfairly pessimistic levels of noise. "

And here's a link: http://www.epanorama.net/links/audiodigital.html

A quick read is saying to me that it does vary from system to system but like I said, I'll need to go over it a couple of times..

quote:

Originally posted by Synchronicity A quick read is saying to me that it does vary from system to system but like I said, I'll need to go over it a couple of times..

Yeah, that's talking about something far more complex based upon an entire system's dynamic range. when you have multiple components connected together... can you have differences in the way the signal flows from one device to another.

I was typing while you posted guys.

So does that mean that the more bits you work at the broader the scale is, rather than the louder you can get?

quote:

Originally posted by Synchronicity I was typing while you posted guys. So does that mean that the more bits you work at the broader the scale is, rather than the louder you can get?

The theory is that yes, with more bits, you have a larger capacity for dynamic range... but not overall volume. AND a much quieter noise floor. so if you can work with more bits, do it!

Yes. More bits means that you can specify the relative volume of a signal more precisely. 0 dB still refers to whatever the maximum is that the unit is capable of.

Software sometimes breaks this rule when using 32-bit floating point processing; Cubase reserves a few MSBs to allow for positive gain, literally going up to 6 dB or thereabouts. However, 0 dB is still the full swing of your sound module/card/etc. and anything higher will clip if re-recorded or saved - it's just the software that's changed the scale.

That Bob Katz book I referenced has a whole chapter about his proposed K system of measuring audio. There are so many different scales now, dbV, dbU, dbFS, etc. meaning 0db isn't the same for every device.

the K system bob describes, is one unified system that every device could work on.. so when intermixing analog and digital equipment, 50K means 50K on both devices, not like now where 0db and 0db on two can mean different voltages. again, a good read.

DigiNut is a genious also.

quote:

Originally posted by CReddick The theory is that yes, with more bits, you have a larger capacity for dynamic range... but not overall volume. AND a much quieter noise floor. so if you can work with more bits, do it!

There must be a huge difference in dynamic range going from 16 bit to 24 then. I work at 24..

quote:

Originally posted by DigiNut Yes. More bits means that you can specify the relative volume of a signal more precisely. 0 dB still refers to whatever the maximum is that the unit is capable of. Software sometimes breaks this rule when using 32-bit floating point processing; Cubase reserves a few MSBs to allow for positive gain, literally going up to 6 dB or thereabouts. However, 0 dB is still the full swing of your sound module/card/etc. and anything higher will clip if re-recorded or saved - it's just the software that's changed the scale.

Wahey! Most Significant Bits, I just learnt that on Tuesday!

We just briefly talked about it, we set up an oscilloscope between two computers and transferred/read ascii characters through the oscilloscope. It was quite cool to see actually.

So with the 32 bit floating point.. since my card supports a maximum depth of 24 does that mean I can't work at the 32 bfp depth?

quote:

Originally posted by CReddick That Bob Katz book I referenced has a whole chapter about his proposed K system of measuring audio. There are so many different scales now, dbV, dbU, dbFS, etc. meaning 0db isn't the same for every device. the K system bob describes, is one unified system that every device could work on.. so when intermixing analog and digital equipment, 50K means 50K on both devices, not like now where 0db and 0db on two can mean different voltages. again, a good read.

Yeah I'll need to take a look at that.

quote:

DigiNut is a genious also.

Hehe, I had a feeling Diginut would be able to answer my question.

quote:

Originally posted by Synchronicity There must be a huge difference in dynamic range going from 16 bit to 24 then. I work at 24..

Probably not super noticalble.

quote:

Originally posted by Synchronicity Wahey! Most Significant Bits, I just learnt that on Tuesday! We just briefly talked about it, we set up an oscilloscope between two computers and transferred/read ascii characters through the oscilloscope. It was quite cool to see actually. So with the 32 bit floating point.. since my card supports a maximum depth of 24 does that mean I can't work at the 32 bfp depth?

The 32-bit is a floating point in software. your sequencer dithers back down to 24 bit before making output. The higher floating point helps out complex calculations. Think of it as more decimal places in a number.

quote:

Originally posted by Synchronicity So with the 32 bit floating point.. since my card supports a maximum depth of 24 does that mean I can't work at the 32 bfp depth?

It's just how the software operates. It has nothing to do with your soundcard. The only thing where the bit depth of your soundcard matters is with playback and recording. You can process and render 24-bit audio even if you only have a 16-bit card, if I'm not mistaken.

quote:

Originally posted by CReddick Probably not super noticalble. The 32-bit is a floating point in software. your sequencer dithers back down to 24 bit before making output. The higher floating point helps out complex calculations. Think of it as more decimal places in a number.

Right. Again that was mentioned at Uni - I usually need to go over these things for them to sink in.

So is 32 bfp much more cpu intensive?

Mmn. I didn't really need to ask that, I've set it up and there's no difference.

I guess a better question is are there any disadvantages to it?

I think every sequencer works with 32-bit floating point internally. The setting you're messing around with is probably a recording setting; when you record you can choose what bit depth you want to record it in and no, there's no real disadvantage to a 32-bit recording except for bandwidth. If you're streaming a bunch of files off a disk, you have to read more bytes at a time if they're 32-bit, which means you can't play back as many at once (someone actually posted about a disk bandwidth problem a few weeks ago).

If you don't do a lot of recording, sampling, etc., then it probably won't make much difference.

If your sequencer is actually letting you change the bit depth of its internal processing then that seems like a bit of a WTF to me... 32 bits is the actual number of bits in a single-precision floating point number on a PC so it would be pretty hard to go below that. Easy to go above, though, to a double (64) or extended (80). Wasteful, though, since the hardware doesn't support more than 24 bits and 32 bits is plenty enough to avoid any rounding errors.

quote:

Originally posted by DigiNut I think every sequencer works with 32-bit floating point internally. The setting you're messing around with is probably a recording setting; when you record you can choose what bit depth you want to record it in and no, there's no real disadvantage to a 32-bit recording except for bandwidth. If you're streaming a bunch of files off a disk, you have to read more bytes at a time if they're 32-bit, which means you can't play back as many at once (someone actually posted about a disk bandwidth problem a few weeks ago). If you don't do a lot of recording, sampling, etc., then it probably won't make much difference. If your sequencer is actually letting you change the bit depth of its internal processing then that seems like a bit of a WTF to me... 32 bits is the actual number of bits in a single-precision floating point number on a PC so it would be pretty hard to go below that. Easy to go above, though, to a double (64) or extended (80). Wasteful, though, since the hardware doesn't support more than 24 bits and 32 bits is plenty enough to avoid any rounding errors.

Yeah actually I'm using SX3 and it's the record format I've changed.

What you've just said leads me to loads of other questions but I'll just let this info sink in for now!

Cheers.

Powered by: vBulletin
Copyright © 2000-2021, Jelsoft Enterprises Ltd.