Re: Aubio down to the sample

[Paul Brossier <piem@altern.org>]

On Sun, Nov 28, 2004 at 05:52:42PM -0800, Tom Smith wrote:
> Paul,
>  
> I'm not sure if you received my previous e-mail, but I'm
> interested in knowing whether aubio can determine the exact
> location of an onset, down to the sample.  I understand there
> must be a latency to accommodate both the forward analysis and
> the onset detector itself, but is there a method to label the
> exact instance the onset occurred?  I've been looking at the
> peak picker code and it appears that it only gets fed a value
> every 512 samples, so it would seem that it has no way of
> knowing when it occurred within this window.  Can you think of
> anything that might resolve the location down to the sample?

Hi,

Oops, no I did not receive your previous email.

Simple things I can think of:

Have the peak picking select the local minima before the actual
peak, (instead of a fixed 3 frames ahead).

Look for the closest zero crossing within the frame given by the
peack picker. (will give nicer cuts, but early or late ones may
still happen).

I will try to have a go with python/aubiocut later this week.

Bye, Paul

ps: mailing copy to the new mailing list aubio@piem.org.

>  
> Thanks again, Tom Smith tsmith32768@yahoo.com
>  


-- 
To unsubscribe, send mail to aubio-unsubscribe@piem.org.