Sensible defaults could make Audacity a lot better

[waxcylinder]

It's in the numbers, Steve. WC acknowledged that when he wrote that most YT video's about Audacity distorted.

So you are suggesting that the linear track view is directly responsible for people recording so loud that their recording is clipped? That's a bit far fetched isn't it?

I don't know about Cyrano bit I'm certainly suggesting that - not far fetched at all

I came to Audacity from the world of cassette decks and on my Nak BX-2 with TDK-SA tapes that it was set for I knew I could successfully push the meters to near the extreme and the tape would "soak it up" (I have a fair grasp of the physics involved).

So when I started out I worked with Audacity with it's default settings (including waveform view) I thought the same applied and I pushed the levels to the max in the waveform (I had a poor grasp of DSP)- and remember the meters were a bit inadequate - that's why I worked on and helped lobby through the proposal that ended up giving us the lovely new meters with the much more useful and usable green-yellow-red colorization.

To the end of trying to help users avoid oversaturating their signal when in default waveform view I still have an outstanding proposal on the Wiki
http://wiki.audacityteam.org/wiki/Propo ... rm_display
This was originally a subtopic in the proposal for the new meters but I dug it out from there to stand alone once we had the new meters - there does seem to be some limited support for this proposal from Gale and James.

If this proposal was implemented than we might not need to consider Cyrano's point 2 in the original post of this thread, to make Waveform (dB) the default view.
I have a sneaking suspicion that changing the default view to dB might unsettle a lot of our users.

WC

[waxcylinder]

However I am broadly in support if cyrano's point 1 - that the data in the Track Control panel should accurately represent the actuality of the track.

If we can get some broad agreement on this I am prepared to hammer out a Wiki proposal ( the fast track )

WC

[steve]

I came to Audacity from the world of cassette decks and on my Nak BX-2 with TDK-SA tapes that it was set for I knew I could successfully push the meters to near the extreme and the tape would "soak it up" (I have a fair grasp of the physics involved).

So when I started out I worked with Audacity with it's default settings (including waveform view) I thought the same applied and I pushed the levels to the max in the waveform (I had a poor grasp of DSP)- and remember the meters were a bit inadequate

So if you had seen this as the track view when making a voice recording, would you have tried to push it higher, or would you have been satisfied that it was high enough?

firsttrack005.png (12.6 KiB) Viewed 836 times

Note that many of the over-saturated recordings on YouTube look something like this in linear waveform view:

firsttrack006.png (11.81 KiB) Viewed 836 times

[steve]

I am broadly in support if cyrano's point 1 - that the data in the Track Control panel should accurately represent the actuality of the track.

So am I, but the point that I have been trying over and over is that the logarithmic view is no more "realistic" than the linear view. They present different visual representations. Each has strengths and weaknesses. Neither of them give a good view of frequency content - for that you need yet another view - "Spectrogram". Neither give a good view of stereo spread - for that you want a "vectorscope display" (ask koz).

If you have highly compressed music (typical of most modern "popular music"), the logarithmic view makes it extremely difficult to tell if the audio is clipped or just highly compressed because the waveform is "squashed up" at the top end. With a linear view it is much easier to see.


Common waveform shapes are visually distorted when scaled logarithmically:

Linear view

tracks013.png (27.93 KiB) Viewed 836 times

logarithmic view

tracks014.png (22.65 KiB) Viewed 836 times

The logarithmic view also gives very poor representation of noise level. Low level noise can look MUCH worse than it really is if it is above the bottom end of the log scale, or will look totally silent if it is below.
See: http://forum.audacityteam.org/viewtopic ... 48#p283348
and: http://forum.audacityteam.org/viewtopic ... 56#p281856


Linear waveform view is "conventional" in that it is frequently the default waveform view in other audio software:
See: http://forum.audacityteam.org/viewtopic ... 72#p283272


In the manual we describe the standard "Fade In" and "Fade Out" effects as "linear fades". This is the correct and ubiquitously accepted terminology for this type of fade, but see here: http://forum.audacityteam.org/viewtopic ... 77#p283277


I am not saying that a logarithmic view does not have its uses - quite the opposite, I am saying that it does have its uses, but it is no more "realistic" or "truthful" than other waveform visualisations. As shown above, in many ways it is more misleading than our current default view.

And please note that I am not talking about the use of "dB". That is just how we label the vertical ruler and is not dependent on whether the waveform is scaled linearly, logarithmically or any other scale.

Arguably, if we want to represent "how it sounds" we should use "sone" scaling (rather tricky to implement as it is frequency dependent, but not impossible).

[Robert J. H.]

(snip)
Arguably, if we want to represent "how it sounds" we should use "sone" scaling (rather tricky to implement as it is frequency dependent, but not impossible).

+1
Another problematic point apart from the frequency is the actual output or target volume.
The Sones will change depending on this setting, -40 dB won't be the same as +80 dB. Thus, yet another control is needed.

Robert

[cyrano]

Arguably, if we want to represent "how it sounds" we should use "sone" scaling (rather tricky to implement as it is frequency dependent, but not impossible).

That would be even better. But I was looking for a low-cost solution to a perceived problem. I don't think implementing "sone" will be low-cost. And that might lower the chance of it happening considerably.

[waxcylinder]

So if you had seen this as the track view when making a voice recording, would you have tried to push it higher, or would you have been satisfied that it was high enough?

firsttrack005.png

Actually I've always worked with and managed level with the meters rather than the waveforms.

First learning was that I could make (even the old) recording meter much bigger (thanks Koz).

Second learning was that -6dB was plenty high enough peak to aim for.

Third learning was that the -6dB peak on the meter is roughly 50% on the linear waveform.

But it did take some learning and a fair while - learning that I have tried to pass on through stuff I have written in the Manual.

WC

[steve]

+1

Is that +1 for having a "sone scaled view" or +1 for a sone scale view being the default view (bearing in mind that much less than 1% of our users are likely to have the slightest idea of what they are looking at)?

[steve]

It's in the numbers, Steve. WC acknowledged that when he wrote that most YT video's about Audacity distorted.

The majority of users are importing music. And that music is usually compressed to death. It's as simple as that.

If that is the primary concern, then I venture to say that a better solution would be for "show clipping" to be enabled by default.
It would be virtually impossible for a sighted user to not notice that there is a problem with this recording:

Audio track with thick blocks of red clipping indicator lines

firsttrack007.png (15.81 KiB) Viewed 833 times

[waxcylinder]

The majority of users are importing music. And that music is usually compressed to death. It's as simple as that.

If that is the primary concern, then I venture to say that a better solution would be for "show clipping" to be enabled by default.
It would be virtually impossible for a sighted user to not notice that there is a problem with this recording: ...

+1 to that