Re: [Audacity-devel] Normalize effect extended

[Dominic M. <do...@au...>]

Markus Meyer wrote:
> Vaughan,
> 
> first, thank you for your clear words. I know that sometimes I can be a
> bit "snappy" (is this the real word for it?) in discussions. Also, my
> incomplete knowledge of the English language does not always allow me to
> phrase things as polite as I should. I hope this is not misunderstood.

I hope there have never been any misunderstandings.  I think that strong 
debate on the list is a good thing, so in general I think it is better 
to speak your mind and explain why you think one thing is a good or bad 
idea, rather than hold back in fear of offending anyone.  Plus, compared 
to many other open-source mailing lists, I think the Audacity lists are 
always very friendly.

> You're right that I'm a proponent for beginners here. Infact, I think
> Audacity is (becoming) the perfect tool for beginners to edit audio.
> This is especially important as there are no real alternatives in the
> "beginner" market. For example, though it is a good thing if a
> professional sound studio can use Audacity for mixing and mastering,
> they will have no problem buying ProTools (or Samplitude or anything
> else). But even if the beginner spends the money to buy
> ProTools/Samplitude/Whatever it doesn't help because it is too
> complicated for casual use. So Audacity is filling a real gap here.
> Thinking about this a bit more, we should really try to come up with
> ways to make audio editing easier. For example, when making a
> presentation with PowerPoint, I can say "make this page slide in from
> the left". I do not have to say, "set the x offset of the main layer to
> -screen_size_x, then start a timer with t=50ms, and everytime the timer
> ticks, add 1 to the x offset of the main layer, then when the x offset
> is at 0 again, stop the timer". But with audio programs I can't say
> "make this vocal sound brighter", or "make this instrument appear in the
> background" or "give this lead guitar solo more pressure". All I can do
> is say is things like "I want to add 3db to all frequencies in the
> 1,5khz to 3khz range", "I want a reverb applied with this-and-that
> predelay" etc. All of this is totally unrelated to the actual tasks I
> want to perform. I'm not saying, a professional shouldn't have access to
> all those parameters. But I shouldn't have to read a course in signal
> processing just to make a vocal sound brighter and more present.

Very much agreed.  However, I would like to say that I think the real 
strength of Audacity is not just that it's "easy" for beginners, but 
that it is easy without trying to "dumb it down".  Therefore while I'm a 
big fan of making it more clear, to disagree with Vaughan I don't think 
that using the word "distorting" instead of "clipping" helps.  I'd 
prefer to have a "?" icon - a context-sensitive help button - next to 
that word and let people click on it to learn what "clipping" means.

More below:

> Sorry for digressing, back to the actual discussion:
> 
> Vaughan Johnson schrieb:
> 
>>    * Move "Remove DC Offset" to Amplify, or to its own single-purpose
>>      effect (somewhat less desirable because the Effects menu is
>>      already long). I know lots of novices wonder what DC Offset is.
>>    * Change Normalize to have no controls other than Preview, Cancel,
>>      and OK buttons, and static text that says "Normalize maximum
>>      amplitude to 0dB", or maybe even "Make selection as loud as
>>      possible without distorting." (avoiding the word "clipping"). It
>>      doesn't need a checkbox because all it can do is Normalize.
> 
> This sounds fine to me, though I think we should not have an Effect
> dialog that just contains Preview/Cancel/OK and nothing else. But we
> could make it behave like the Fade-In/Fade-Out effects, it would just be
> applied without showing a dialog at all.
> 
> I also think that it is a good idea to always apply the DC offset when
> normalizing (Martyn suggested that). I'm not familiar with the DC offset
> removal algorithm. Is this something that "just works" or is it a
> heuristic algorithm than can easily fail on certain input data? If the
> latter, we shouldn't perform it automatically everytime we do a normalize.

Consider also that I'm hoping to move in the direction of support for 
real-time effects.  In that case there is a more clear distinction: to 
normalize and remove DC offset requires analyzing the entire audio, and 
can't be done in real-time.  Amplify, though, doesn't require any 
pre-analysis and could be real-time.

- Dominic