Re: [Audacity-nyquist] Normalizing long tracks

[Stevethefiddle <ste...@gm...>]

Thanks Edgar,

Apologies for bringing up a subject that has already been discussed many 
times, but searching audacity-nyquist on Nabble also brought a plentiful 
supply of posts that suggest it could be done.

I've now found this post that you wrote almost exactly a year ago 
http://n2.nabble.com/Nyquist-in-Audacity-crashes-with-long-sound-files-td238715.html#a238715

This usefully explains why it doesn't work, and also gives a strong hint 
as to why things like (snd-avg ...) and (extract ... ) exhibit the same 
limitation.
There has been some progress in the past year relating to this -  
'autonorm', 's-read' and 's-write' are now implemented in Audacity. 
Unfortunately the memory limitation of "s" being a globally bound 
variable is still with us.

Funily enough, I remember reading that post last year but at the time I 
was totally new to Nyquist and did not understand it. Reading it again 
it looks like the limitation in Audacity is inevitable. Unfortunately 
even (play ...) allocate huge amounts of memory, so I can't even get 
round the problem by writing the audio to disk.

Thanks very much for the reply - now that I have some understanding of 
why the problem exists I'll just work around it. Probably the easiest 
way is to disable normalizing with something like: (if (> LEN 1000000) 
..... )

Steve D


edgar-rft [via Audacity] wrote:
> Steve wrote:
>
>  > As much as I dislike bumping posts, can very long tracks be
>  > Normalized with Nyquist without having to write the file to disk?
>
> This topic has been discussed at least twenty times (or even more)
> on this list in the last few years and AFAIK it's still not possible
> to use sample-evaluating functions like PEAK, SND-FETCH, etc. in
> Audacity without causing huge memory allocation.
>
> This problem is probably still unsolved.
>
> - edgar


-- 
View this message in context: http://n2.nabble.com/Normalizing-long-tracks-tp4463058p4482925.html
Sent from the audacity-nyquist mailing list archive at Nabble.com.