<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'>Thanks for the change notes and other remarks! Indeed, by manipulating the parameters you mentioned I could tweak the onset detection algorithm to behave as in the alpha of aubio. Makes things easier if I can just recommend users to grab latest released version of aubio.<br><br>Cheers,<br>Lucas<br><br><div>> Date: Wed, 19 Mar 2014 18:05:50 -0300<br>> From: piem@piem.org<br>> To: lukasz.tracewski@outlook.com; aubio-user@aubio.org<br>> Subject: Re: [aubio-user] Onset detection in Python<br>> <br>> On 03/19/2014 05:36 PM, Lukasz Tracewski wrote:<br>> >> This is fascinating. Would you mind if I prepare some short post<br>> >> to publish on aubio's blog?<br>> ><br>> > Not at all, I will be honoured!<br>> ><br>> > If you would like to know more about the project, I made a small<br>> > write-up in PDF available here:<br>> ><br>> https://github.com/tracek/Ornithokrites/blob/master/report/report.pdf?raw=true<br>> ><br>> > The project's development page: https://github.com/tracek/Ornithokrites/<br>> <br>> excellent, I'll prepare a short post and submit it to you before publishing.<br>> <br>> > One more question related to aubio (energy) onset detection. In my<br>> > early development version I was using 0.4.0alpha. Now that I switched<br>> > to 0.4.1 some more onsets gets detected, in this case false<br>> > positives. Did the algorithm get more sensitive? I tried tuning the<br>> > threshold but no avail. I did not notice anything particular in the<br>> > release notes. In the history of onsets you mention cleaning up and<br>> > there is also this change:<br>> ><br>> > aubio_onset_set_delay_ms(o, 50.); to: aubio_onset_set_delay(o, 4.3 *<br>> > hop_size);<br>> ><br>> > So making delay dependent on hop size. I tried compiling with old<br>> > code, but actually got even more different results. Are there any<br>> > fixes that could concern energy onset detection between 0.4.0alpha<br>> > and 0.4.1? I was contemplating going back to alpha, but that is<br>> > usually not a smart idea (probably it is just coincidence that alpha<br>> > works temporarily better for my training set, perhaps a matter of<br>> > tuning some other parameters than threshold).<br>> <br>> I'm not sure exactly what version you had before. 'alpha' is a moving<br>> target: right after a release, the version number is increased, and<br>> ~alpha added to it. (feel free to send me a tarball of the source code<br>> you used for 0.4.0alpha if you want me to look at it more in details).<br>> <br>> as far as onsets are concerns, only two changes between 0.4.0 and 0.4.1<br>> should affect the results:<br>> <br>> - i changed back the delay from 50ms to 4.3 * hop_size, because the<br>> peak-picking algorithms depends solely on the hop_size, not on the<br>> temporal resolution of the novelty function. the default value can not<br>> be set at a fixed time.<br>> - i also changed the changed the minimum inter-onset-interval (minioi)<br>> from 5 * hop_size to 20ms, so that could explain some more of the<br>> changes you observed.<br>> <br>> now you can set the delay and minimum IOI yourself. from python, you can<br>> set both these values using the following functions:<br>> <br>> - aubio.onset.set_delay, aubio.onset.set_delay_s, and<br>> aubio.onset.set_delay_ms<br>> - aubio.onset.set_minioi, aubio.onset.set_minioi_s, and<br>> aubio.onset_set_minioi_ms<br>> <br>> you can also get the current value of the onset object with the<br>> corresponding getters.<br>> <br>> see the doc here: http://aubio.org/doc/latest/onset_8h.html<br>> <br>> cheers, Paul<br>> <br>> <br></div> </div></body>
</html>