Contrast Analysis not Working

[jasperjones]

Problem occurs in 2.0 on Win 7 Pro x64 installed with the .exe installer.

The thread title basically says it all. I frequently used contrast analysis in the beta, now it's not working anymore. Regardless of what I do, whenever I click the "Measure selection" button in the Contrast Analysis box, I get an error message "you can only measure on track at a time" (yes, i know) and a contrast analysis result of "Please enter valid times." Needless to say, I do enter valid times.

Thoughts?

[steve]

The Contrast tool only works on mono tracks.
I'm unsure if the tool was ever intended to work on stereo tracks. Did it work on stereo tracks in your previous version? Do you know the exact version number that you were using previously?

[Gale Andrews]

The Contrast tool only works on mono tracks.
I'm unsure if the tool was ever intended to work on stereo tracks.

Previously, Contrast measured the bottom track in the project irrespective of which track was selected. If you only had one stereo track, Contrast would just analyse the bottom (right) channel. If you had five stereo tracks, it analyzed the right channel of the fifth stereo track.

Contrast was then fixed so that it only measured a single selected mono track.

So to measure contrast on a stereo track now, use the Track Drop-Down Menu to "Split Stereo Track", then you can select and analyse one channel of the track or the other.



Gale

[jasperjones]

Thank you very much!

[ahjteam]

The Contrast tool only works on mono tracks.
I'm unsure if the tool was ever intended to work on stereo tracks.

The Manual says:
http://manual.audacityteam.org/man/Analyze_Menu

"Contrast Analysis is designed to analyze a single mono or stereo speech track"


So if it would be possible, could someone make it work so that it works again on stereo tracks (both Windows and Mac and Linux versions)?

[Gale Andrews]

The Contrast tool only works on mono tracks.
I'm unsure if the tool was ever intended to work on stereo tracks.

The Manual says:
http://manual.audacityteam.org/man/Analyze_Menu

"Contrast Analysis is designed to analyze a single mono or stereo speech track"

Thanks for the reminder that we did not update the Manual, which I have now done. The Manual was wrong in suggesting that "if you have multiple tracks open, Contrast Analysis will calculate rms values based on all of the tracks, whether they are selected or not" - as above, it analysed only the bottom track or channel.

So if it would be possible, could someone make it work so that it works again on stereo tracks (both Windows and Mac and Linux versions)?

As above, it never did analyze a stereo track, only the bottom (right) channel of it. Averaging out rms in multiple tracks would mean more coding and complexity which might not be worth it. And if the user has selected multiple tracks, it probably isn't useful for Contrast to make a decision to operate just on the bottom one.

I agree it would save user's time that if an unsplit stereo track was selected, Contrast simply went ahead and analyzed the right channel with an appropriate message. But again this may mean significant new coding.



Gale

[ahjteam]

The Contrast tool only works on mono tracks.
I'm unsure if the tool was ever intended to work on stereo tracks.

The Manual says:
http://manual.audacityteam.org/man/Analyze_Menu

"Contrast Analysis is designed to analyze a single mono or stereo speech track"

Thanks for the reminder that we did not update the Manual, which I have now done. The Manual was wrong in suggesting that "if you have multiple tracks open, Contrast Analysis will calculate rms values based on all of the tracks, whether they are selected or not" - as above, it analysed only the bottom track or channel.

So if it would be possible, could someone make it work so that it works again on stereo tracks (both Windows and Mac and Linux versions)?

As above, it never did analyze a stereo track, only the bottom (right) channel of it. Averaging out rms in multiple tracks would mean more coding and complexity which might not be worth it. And if the user has selected multiple tracks, it probably isn't useful for Contrast to make a decision to operate just on the bottom one.

I agree it would save user's time that if an unsplit stereo track was selected, Contrast simply went ahead and analyzed the right channel with an appropriate message. But again this may mean significant new coding.



Gale

Hey Gale. Can you change it so, that it calculates both channels (so just does two passes, first left and then right), then just adds them together and divides the result by 2 for an average? (a+b)/2

[steve]

Hey Gale. Can you change it so, that it calculates both channels (so just does two passes, first left and then right), then just adds them together and divides the result by 2 for an average? (a+b)/2

That probably isn't the correct way to handle stereo tracks, but more importantly the WCAG 2.0, Success Criteria 1.4.7 does not specify how stereo sound should be handled. http://www.eramp.com/WCAG_2_audio_contr ... l_help.htm

For example, if a voice is panned to one side and background music is panned to the other side, then it will probably be easier to pick out what the voice is saying (less contrast) than if both voice and music are panned centre.

(The other issue is that neither Gale nor myself are Audacity developers.)

[Gale Andrews]

For example, if a voice is panned to one side and background music is panned to the other side, then it will probably be easier to pick out what the voice is saying (less contrast) than if both voice and music are panned centre.

Yes, so if you split the stereo track then run Contrast on each channel, it isn't easy to interpret the result.

It's also unfortunate that for most of the examples WCAG actually show a selection in a stereo pair, which now won't work and which scenario wasn't considered by them.

Gale

[steve]

It's also unfortunate that for most of the examples WCAG actually show a selection in a stereo pair, which now won't work and which scenario wasn't considered by them.

So it appears that they were working on the assumption that stereo tracks are fairly similar in both channels?

If that's the case, what was the reason for the "fix"?