< Back to IRCAM Forum

Onseg/Chop not segmenting properly when replacing corpus in MuBu mosaicing example

hi guys

I was trying to get into mosaicing via MuBu and would like to use it for recorded speech samples.

my first approach was to look into the included example (mubu-howto-temporal-modeling) and simply replace the Freesound drumloops with random speech .wav files that I found on youtube.

however, something strange happened: if I include those into the Corpus, about 90% of the time the chop and onseg slices are not calculated properly anymore. when I replace and reload the corpus with the old drum samples it still won’t work.

only when I restart the whole patch with the original corpus it seems to work again.

this may be really basic but I was still wondering if anyone could explain it to me?

I can provide my samples and any steps to reproduce more exactly if unclear.


Hi Gabriel,
I see, the mubu.process doesn’t go to the end sometimes (first outlet stops before 1.), just bang until it does, until we find a fix…
With your own sounds, also try different onseg.threshold values.

hi Diemo
ok thanks for getting back about this - I was really irritated because I couldn’t figure it out.