[mirtoolbox] Onsets Question...
- From: Ahmed Nagi <ahmed.nagi@xxxxxxxxx>
- To: mirtoolbox@xxxxxxxxxxxxx
- Date: Thu, 6 May 2010 17:14:47 +0200
Hello....
I hope this is the correct place and way to ask a question regarding the MIR
toolbox.
I was reading the documentation for the most recent version 1.2.5, and i have
noticed an update that confuses me a bit...
The situation arise when i am using the mironsets(Filename, 'Envelope'),
without specifying further option.
* On page 80, it reads that the 'Spectro' option is the default
* yet on page 82, it reads that " the audio waveform is by default first
decomposed into channels (cf. the ‘Filterbank’ option below) ".
So if i am writing mironsets(Filename, 'Envelope), what is the order of
operations that is being performed:
a) The spectral energy (power) is computed directlty from the signal, from
which an envelope (using hanning window, and some setting for the window size,
hop size) is extracted to compute the onsets
OR
b) the signal is first decomposed into channels, and then, FOR EACH channel, an
envelope is computed using the spectral energy, and the curve are summed
together, to be passed the peaks detection,....
Thank you
Ahmed
Other related posts:
