Audacity Forum

cyrano wrote: There isn't any sound engineer on the planet that thinks in linear percentages.

I agree. That's why I would like an option for a dB scale on the linear waveform view.
Most of our users are not sound engineers and come to Audacity with little idea of what "dB" means.

Which of these would you call a "linear fade out"?

steve wrote:

cyrano wrote: There isn't any sound engineer on the planet that thinks in linear percentages.

I agree. That's why I would like an option for a dB scale on the linear waveform view.
Most of our users are not sound engineers and come to Audacity with little idea of what "dB" means.

That's why we need to educate them. There is no place on the web, no book about audio or any other resource that doesn't use dB's.

You can adjust to the user's ignorance or you can try to educate them. Adjusting to ignorance is just plain wrong. They'll never learn and will continue to have problems with the rest of the world.

steve wrote:Which of these would you call a "linear fade out"?

None of these. A linear fade out is perceived by the human ear. That is logarithmic when it comes to the perceived loudness for one isolated tone. Once you apply it to music (or even voice) that behaviour changes according to the Munson curve. All implementations of fade out are approximations. None are perfect. Thus it doesn't really matter a lot.

steve wrote:Practical use case for 32-bit float:
Mid production, I want to make lossless backups of my valuable tracks. I export as 32-bit float WAV.

Practical case 2:
Impulse files for convolution effects.

Practical use case 3: we actually recommend its use for archiving/backup a high quality copy of the raw capture in this workflow tutorial in the Manual

http://manual.audacityteam.org/o/man/sa ... n.html#raw

WC

steve wrote:

cyrano wrote:Yet, Audacity shows 32 bit in the left pane of the main window. That is extremely confusing.

I agree, I think that should be hidden from users, because: (a) it is (as you say) confusing. (b) may not actually be true.

If that is indeed the case, does this feature request boil down to removing the text that says "32-bit float" from the Track Control Panel?

And while we are about it should we also be considering removing from there also the text that says "44100 Hz"?

WC

cyrano wrote:

steve wrote:

cyrano wrote: There isn't any sound engineer on the planet that thinks in linear percentages.

I agree. That's why I would like an option for a dB scale on the linear waveform view.
Most of our users are not sound engineers and come to Audacity with little idea of what "dB" means.

That's why we need to educate them. There is no place on the web, no book about audio or any other resource that doesn't use dB's.

You can adjust to the user's ignorance or you can try to educate them. Adjusting to ignorance is just plain wrong. They'll never learn and will continue to have problems with the rest of the world.

So, can you tell me what's the logarithm of -0.5?

The only right way to display a wave form is the linear scale - it represents voltage.
Thus, you have to square all sample values before you can even attempt apply a dB scale - and lose some information.
It doesn't count what a sound engineer thinks but what's physically correct.

Robert

waxcylinder wrote:

steve wrote:

cyrano wrote:Yet, Audacity shows 32 bit in the left pane of the main window. That is extremely confusing.

I agree, I think that should be hidden from users, because: (a) it is (as you say) confusing. (b) may not actually be true.

If that is indeed the case, does this feature request boil down to removing the text that says "32-bit float" from the Track Control Panel?

And while we are about it should we also be considering removing from there also the text that says "44100 Hz"?

Have I stated that?

I'd like to see REAL, accurate values. And 32 bit is only an internal format. It doesn't have anything to do with the quality of the source and its ALWAYS 32 bit, so it offer NO real information.

Worse, it's plainly wrong in some cases.

Robert J. H. wrote:

cyrano wrote:

steve wrote:

cyrano wrote: There isn't any sound engineer on the planet that thinks in linear percentages.

I agree. That's why I would like an option for a dB scale on the linear waveform view.
Most of our users are not sound engineers and come to Audacity with little idea of what "dB" means.

That's why we need to educate them. There is no place on the web, no book about audio or any other resource that doesn't use dB's.

You can adjust to the user's ignorance or you can try to educate them. Adjusting to ignorance is just plain wrong. They'll never learn and will continue to have problems with the rest of the world.

So, can you tell me what's the logarithm of -0.5?

Yes, I could, but what would be the point?

The only right way to display a wave form is the linear scale - it represents voltage.

I'd agree if we were talking about electronics. We are not. And even in electronics, dB is used quite a bit. Nobody would specify a filter as "50% at 300 Hz", fi.

Thus, you have to square all sample values before you can even attempt apply a dB scale - and lose some information.

I'm not sure what you're getting at. Squaring looses information? That's some strange math, there.

It doesn't count what a sound engineer thinks but what's physically correct.

You should go tell that to the EBU, ACX or AES. Should get some laughs...

I'm not at all too sure if discussing is a favorite sport here. I went over ALL the feature requests on this forum and most seem to result in some besides-the-point discussion ending in nothing at all besides some ego lifting.

But hey, if that's the way you like it...

cyrano wrote:Yes, I could, but what would be the point?

It would illustrate that the so called "logarithmic" view does not accurately represent the log sample values.
The other thing that cannot be represented with a logarithmic vertical scale is silence.
http://www.rapidtables.com/math/algebra ... m_of_0.htm
We work around this limitation by ignoring values below a specified level, and by plotting the log of the inverse of negative sample values. The linear scale does not require these esoteric tricks because all valid sample values may be accurately plotted,

cyrano wrote: Squaring looses information? That's some strange math, there.

Not strange math, basic physics. By squaring, you loose phase information - the square of a signal is the same as the square of the inverse signal, but clearly a signal and its inverse are very different things (try mixing a signal with its inverse).

cyrano wrote:I'd like to see REAL, accurate values. And 32 bit is only an internal format. It doesn't have anything to do with the quality of the source and its ALWAYS 32 bit, so it offer NO real information.

Try this:

1) Take three audio tracks and normalize to -3 dB
2) Make one of the tracks 16-bit, one of them 24-bit and one of them 32-bit float (track dropdown menu > Set Sample Format)
3) Amplify all three tracks by +20 dB (you will need to enable "Allow clipping" in the Amplify effect)
4) Amplify all three tracks by -20 dB.

You will immediately see that the track format can be extremely important.

cyrano, you clearly prefer the Waveform (dB) view to the linear Waveform view. You know that you can set that as the default view?
"Edit menu > Preferences > Tracks > Desfault view mode"

We know that some users prefer some defaults over what other users prefer, That's why we have "preference" settings.