Re: [Audacity-quality] New Truncate Silence effect

[Gale A. <ga...@au...>]

| From Steve the Fiddle <ste...@gm...> 
| Mon, 16 Dec 2013 23:56:07 +0000
| Subject: [Audacity-quality] New Truncate Silence effect
> I'd like to propose a replacement for the Truncate Silence effect (yes
> another one).
> 
> In the light of feedback from the current version I have aimed to make
> this effect easy to use for novice users.
> 
> The main differences which I hope will make it more usable are:
> 
> 1) New "easy mode".
> Rather than the "all in one" approach, this effect provides a
> multi-choice menu to select what you want it to do to the detected
> silences. The default "action" is to simply "Truncate" the detected
> silence to a specified duration.
> 
> 2) Easier / more familiar "units".
> * Novice are likely to be more familiar with the concept of "0.5
> seconds" than "500 milliseconds", so durations are set in seconds.
> * Novices are likely to be more familiar with "25 percent" than "4:1",
> so compression is set as a percentage.
> 
> 3) Easier default values.
> * Ignored silence: 0.5 seconds
> * Truncate to: 0.5 seconds
> * Compress to 50 percent (disabled by default)
> 
> 4) Unused controls are greyed out according to the selected "action"
> rather than requiring the user to set "special values" to disable
> truncation or compression.
> * When the default "Truncation" action is selected, the "compression"
> settings are greyed out.
> 
> 5) Immediate user feedback for invalid input via in-line warnings.
> 
> Slightly more advanced users may prefer to use "compression" rather
> than "truncation", so this is  the second option in the "action" menu.
> * When this action is selected, the "Truncation" settings are greyed out.
> * Compression occurs in the same way as the current effect - Silence
> "beyond" the minimum silence is compressed.
> Example:
> Initial silence = 5 seconds
> Ignored Silence = 2 seconds
> Compress to: = 50 %
> The initial silence is 3 seconds over the "Ignored" silence, so the
> "extra" 3 seconds is compressed to 50% of its original length = 1.5
> seconds.
> The output silence will have the "ignored" 2 seconds and the
> compressed 1.5 seconds, so the processed silence will be 3.5 seconds.
> 
> 
> To avoid regression against the current effect, the final "action"
> choice (for 'advanced' users) is the same as the current effect. The
> output is the minimum duration of the two types of silence shortening.
> Though  I have reservations about this "advanced" mode, it
> accomplishes two objectives:
> 1) Regression is avoided.
> 2) It is more "discoverable" because "Truncation" and "Compression"
> can be experimented with independently, and this "advanced" mode is
> simply "whichever produces the shorter silence".
> 
> 
> Other enhancements:
> 1) The -80 dB setting works.
> 2) Silences can be truncated to zero.
> 3) Silences can be compressed to zero.
> 4) Enhanced in-line warnings (exceeding the min/max for any control
> gives a meaningful warning).
> 5) Less restrictive control ranges (durations up to 10,000 seconds,
> accurate to nearest sample at 44100 Hz).
> 
> The attached patch has only been tested on Linux so there may be
> cross-platform quirks to iron out (hopefully not too many :-)
> 
> Steve


Thanks, Steve. It built fine on Windows for me. I've not tested
it much, but is there enough energy/consensus to decide which 
of the three action choices we want and get this into 2.0.6? 

I would much prefer that to releasing the currently committed  
"half-improvement" in 2.0.6. I don't mean that to be derogatory 
at all, but Steve's new version will be much more user-friendly than
the hard to understand and hard to label effect we have now.

Can we do any conversion from ms to seconds in the .cfg settings
from previous Truncate Silence? Otherwise users' previous 
"InitialAllowedSilentMs" and "LongestAllowedSilentMs" settings 
will be disregarded.  
 
I think it may be a little hard to understand the difference between
truncation and compression. The wording "For silence beyond" in 
"Compression" seems to imply that it outputs the result of compressing
that part of the audio beyond the ignore length, but it adds the ignore
length back to that result. 

I'd have no objection if it only output the compressed result. It
would be more "different" to truncation if it was more "aggressive"
in its compression. 

If we leave the Compression action unchanged, I would tend to 
remove "Compress Silence" (that greys out "Truncation") and 
just have the "Minimum of...".  

FWIW The "Truncate Silence" action has an unwanted colon.




Gale