automatically find silent areas in an audio file (spoken voice)
Hi everyone!
Could someone please suggest me how to make Max analyze an audio file (human voice), know which are the silent zones (where they start and where they end). This gives me the ability to jump from one zone to another (the goal is to mix the words randomly during playback).
Thaks
P.
you can dump sample values using peek~ and trigger when level gets higher and lower than
threshold.
that can make a list with onsets or build start-stop slices
Or check flucoma package, which offers something like that,
even that it is somewhat confusing when you look at it for the first time
Did that with flucoma for many systems...
Currently working on something similar, for the record:
Thank you so much for the answers.
I actually need to make a map of the audio file and know where one word ends and another begins, all this automatically, without playing the audio file.
For example the first word starts at 200 milliseconds, ends at 2000 milliseconds, the second word starts at 2600 milliseconds and ends at 4900 milliseconds and so on. I would like to extract this kind of data from the audio file, somehow, and then use this list to jump from one word to another.
FluCoMa has objects that can slice or segment audio files loaded into buffers.
Look at fluid.bufampslice~ to start. The ‘but ’ indicates the object operates on already recorded audio.
These kind of things are highly dependent on the audio and the parameter settings so it may take some tweaking to get things working as you want. FluCoMa has good documentation and examples though.
Actually, there is a usable example of this, not polyphonic, with just one play~, that work like this. The input is an audio file. But can easily be modified to live recording in tt t same buffer.
In case you need plain max objects example, post.
thanks for the tips, I'll definitely checkout fluid.bufampslice~