bouldersoundguy
Well-known member
As I recall, Cool Edit had a particularly excellent waveform display that included a representation of what a converter would do, including converter ballistics.
As I recall, Cool Edit had a particularly excellent waveform display that included a representation of what a converter would do, including converter ballistics.
Ethan Winer
Acoustics Expert
I never used Cool Edit but others offered some ideas. I can tell you for certain that bit depth affects only the noise floor. This is scientifically proven fact that is not up for debate.
--Ethan
snow lizard
Dedicated Slacker
Well it isn't hard to figure out but if you want alternate methodology try this:
1. Generate a sine wave on your computer. I used 440 HZ, probably isn't too critical.
2. Print it at -24 dBfs
3. Save it as a 16 bit file.
4. Attenuate it by 60 dB
5. Save the attenuated version as a different file.
6. Close all files and your software. You've flushed all the caches, you're not in 32 bit float or something, you've rendered the file to the destination format. The damage is done.
7. Open the attenuated file and amplify it by 60 dB.
8. Play the file. Compare it to the one you didn't attenuate.
If it's a sine wave bathed in static you have dithering enabled. If it's a square wave, there's no dithering. If you want to try it without dithering to find the point where it starts to truncate (probably around -60 dB or so) have fun.
Try it. Or not. I don't care!
I didn't just invent this. If you want to learn about it yourself, try a few basic links - it's not top secret information although a lot of the more detailed resources are full of scary looking math.
Audio bit depth - Wikipedia, the free encyclopedia
Quantization (sound processing) - Wikipedia, the free encyclopedia
Analog-to-digital converter - Wikipedia, the free encyclopedia
The last link goes into a bit of detail about other converter issues but covers quantization pretty well.
snow lizard
Dedicated Slacker
I never used Cool Edit but others offered some ideas. I can tell you for certain that bit depth affects only the noise floor. This is scientifically proven fact that is not up for debate.
--Ethan
I don't mean to refute you my friend but if you insist on correcting me, so is quantization error!
M
mfratus2001
New member
Perhaps this will help:
8 bit recording is like drawing with a large marker.
16 bit recording is like drawing with a sharp-point marker.
24 bit recording is like drawing with a fine-tip pen.
For the same width of a line, it takes more strokes of a 24-bit pen to draw the same line as a marker. But the ability to do finely detailed drawings is much better with a 24-bit pen.
8-bit sound sounds "grainier" because it is. Converting 8-bit to 24-bit does not improve the sound any, but converting from 24-bit to 8-bit destroys detail.
8 bit recording is like drawing with a large marker.
16 bit recording is like drawing with a sharp-point marker.
24 bit recording is like drawing with a fine-tip pen.
For the same width of a line, it takes more strokes of a 24-bit pen to draw the same line as a marker. But the ability to do finely detailed drawings is much better with a 24-bit pen.
8-bit sound sounds "grainier" because it is. Converting 8-bit to 24-bit does not improve the sound any, but converting from 24-bit to 8-bit destroys detail.
Last edited:
bouldersoundguy
Well-known member
The factual errors in the above outnumber the spelling errors.
M
mfratus2001
New member
Well, your criticism was accurate, in that I had spelling errors. What "facts" are incorrect? (IYHO)
Any other opinions? I think its an suitable analogy.
(IMHO) Converting from 8-bit to 24-bit does not make your original audio better (you cannot create what was not there) but it does give you a larger "playing field" if you want to filter the source audio in some way.
Is that better?
Last edited:
bouldersoundguy
Well-known member
As has been discussed, more bits give you lower noise. It does not increase the "resolution" or number of bits that represent a given change of voltage.
Converting from 8 to 24 bits may make some processing better, but most DAWs convert files to 32 bit float anyway so there's not much point converting the file beforehand. But the 8 bit noise floor will still be just as high.
Sorry about the spelling jab. I must have been in a mood.
M
mixsit
Well-known member
I would think your original here is not too far off. (but perhaps technically some of it is..
I have seen the differences in the fidelity referenced as in terms of '% of error(s), ie, % of noise/distortion.
To which... About twenty..(er.. ok) 30
years ago I did a little experiment with my first DA30 dat to see what I could learn about what '16bit depth' meant. I purposefully finagled an mic'd acoustic guitar track way- way down low into the converter, then gained platback way back up for a listen.To this day I'm not sure how to interpret this- but (still) curious to me setting somewhat above a bunch of nasty uncorrelated- and IIRC correlated noise and distortions -was the fairly detailed! sound of the ac guitar.
What do you make of that?
M
mfratus2001
New member
I think 16-bits is still plenty of bits for good music, if you aren't too picky. But reduce that to 8-bit, or even 4-bit, and you will hear some major loss of quality, with the volume level kept the same.
I was just making a connection between audio bit depth and the visual world.
But that's just how I think. Sorry if it was off-base.
Happy Holidays, anyway, everybody.
I was just making a connection between audio bit depth and the visual world.
But that's just how I think. Sorry if it was off-base.
Happy Holidays, anyway, everybody.