Re: [Audacity-nyquist] Re: selection for analyzer plugs

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

Domonkos,

Yes, I'm here on this list.

Do I understand you correctly, that your problem is limited to 
sub-sample positioning?  That's a level of detail I hadn't considered.

I agree that it would be relatively easy to just add an offset 
corresponding to the difference between the selection start, and the 
start of the first sample.

However, could you clarify what sort of applications actually need 
label positioning more accurate than +/- one sample?  Or are you 
applying your Nyquist plug-in to a low-sample-rate control envelope or 
something like that?

By the way, I'm also in favor of passing more information from Audacity 
to the plug-in via variables - that way the plug-in could at least have 
all of the information available to it if it wants to.  So even when 
Audacity doesn't do the right thing, you could attempt to correct for 
it with a clever plug-in.

- Dominic

On Nov 24, 2004, at 3:32 AM, Domonkos Nagy wrote:

> Roger,
>
> I agree with Dominic's solution that sounds start at zero, it is clear
> design, the problem is however when you zoom in to the selection to
> a level where you can see individual samples then although nyquist gets
> 0.0 as the start of the sound, however when this 0.0 is returned from
> nyquist for the label track this 0.0 refers to the start of the 
> selection,
> which is in between the first selected sample and the first "unselected
> one".
>
> an ascii drawing for digest receivers (David, apologies, it probably 
> won't
> work with a screen reader):
>
> asterices are the samples, '+' indicates selection boundaries):
>
>         1       *
> *       *               *
>                                 *
> ----+-----------------------+------
>     ^2                      ^
>     selection start         selection end
>
> Nyquist gets the first sample (marked with 1) at 0.0. However if you 
> return
> a list of (time, label) pairs for the label track, 0.0 will be shown
> at selection start (marked with 2).
>
> It can be confirmed with the following nyquist plug (zoom in enough to 
> see
> individual samples and select just a few of them):
>
> ;nyquist plug-in
> ;version 1
> ;type analyze
> ;name "Mark individual samples"
> ;control debug "debug" int "%" 0 0 0
> ;author "Domonkos Nagy <na...@ya...>"
>
> (setf s1 (if (arrayp s) (snd-add (aref s 0) (aref s 1)) s))
>
> (setf sample-rate (snd-srate s1))
> (setf time-increment (/ 1 sample-rate))
>
> (do ((c 0.0)
>      (l NIL)
>      (curr (snd-fetch s1)))
>     ((not curr) (print l) l)
>   (setq l (cons (list c "S") l))
>   (setq c (+ c time-increment))
>   (setq curr (snd-fetch s1)))
>
> I hope this make sense now. I think a good solution would be to start
> selection on the first sample itself which would fix this problem and
> would be consistent with "select-all" too.
>
> BTW is Dominic (or one of the main developers) following this list at 
> all?
>
> best,
> Domonkos
>
> On Tue, Nov 23, 2004 at 01:58:02PM -0500, Roger B. Dannenberg wrote:
>> I get the digested version of this list, and the images and sounds 
>> Domonkos
>> included arrived here as hex rather than attachments I could easily 
>> view.
>> Anyway, I gather the problem is that the plug-in does not get 
>> information
>> about the location of the selection.
>>
>> I'm not sure what is the best solution. Nyquist sounds are more than 
>> just
>> samples; they contain a scale factor, a start time, a logical stop 
>> time (the
>> start time of the next element in a sequence), and a physical stop 
>> time
>> (after this point, all samples are zero). It would make sense for 
>> Audacity
>> to pass the selection to Nyquist with a start time that reflects the 
>> offset
>> from the beginning of the file. It would also make sense to set up the
>> Nyquist environment to indicate the start time of the selection (see 
>> the
>> AT-ABS behavior). I think Dominic and I talked about these 
>> possibilities
>> long ago and Dominic wanted to keep things simpler by just having 
>> sounds
>> start at time zero. I'll be the first to admit that while Nyquist 
>> does a
>> pretty good job of doing the "right thing" with all these time 
>> offsets and
>> transformations, it can be very confusing and counterintuitive. Since 
>> people
>> usually test Nyquist code without time offsets, it's easy to create 
>> bugs
>> that don't show up until you test it under a time transformation. A 
>> lot of
>> users might be confused when their code stops working when applied to 
>> a
>> selection because the selection runs their code with a time offset or
>> stretch factor corresponding to the selection.
>>
>> An easy solution is simply to pass in selection information by 
>> setting some
>> globals and let the plug-in deal with it. This would inelegantly 
>> bypass some
>> nifty features in Nyquist already set up to handle such problems, but 
>> on the
>> other hand, it might make life easier overall.
>>
>> This is probably at least a good time to revisit the original design 
>> and
>> make sure it's still a good idea we want to perpetuate.
>>
>> 	-Roger
>>
>>
>>
>>
>>
>> -------------------------------------------------------
>> SF email is sponsored by - The IT Product Guide
>> Read honest & candid reviews on hundreds of IT Products from real 
>> users.
>> Discover which products truly live up to the hype. Start reading now.
>> http://productguide.itmanagersjournal.com/
>> _______________________________________________
>> Audacity-nyquist mailing list
>> Aud...@li...
>> https://lists.sourceforge.net/lists/listinfo/audacity-nyquist
>
>
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real 
> users.
> Discover which products truly live up to the hype. Start reading now.
> http://productguide.itmanagersjournal.com/
> _______________________________________________
> Audacity-nyquist mailing list
> Aud...@li...
> https://lists.sourceforge.net/lists/listinfo/audacity-nyquist