Jean-Philippe Jullin

I'm working on a simple AI, analyzing example sounds to then try to reproduce them. My first step is to analyse the audio examples, and I am using the pipo~ module from the MuBu Package for it. It works quite well, but the thing is I have to wait for the entire sounds to be played to get the informations I need (I'm talking about audio descriptors, such as frequency, energy, periodicity, AC1... as many as possible). 

My question is: is there any way to get the same informations while playing the examples faster? 

I'm talking about playing them *5 or *10 their original speed, but right now a simple *2 gives me completely false results. I've attached a simple example patch to show what I am trying to achieve. I used the mean object for the simplicity of the example, but I would need all the numbers coming out of pipo~ (I use them to control some markov chains). 

I'm not sure pipo~ is really meant for such precision, maybe there are other objects more suitable for this?

comment

message

button

mean

pipo~

ezdac~

play~

buffer~

Hi everyone,

I'm working on a simple AI, analyzing example sounds to then try to reproduce them. My first step is to analyse the audio examples, and I am using the pipo~ module from the MuBu Package for it. It works quite well, but the thing is I have to wait for the entire sounds to be played to get the informations I need (I'm talking about audio descriptors, such as frequency, energy, periodicity, AC1... as many as possible). 

My question is: is there any way to get the same informations while playing the examples faster? 

I'm talking about playing them *5 or *10 their original speed, but right now a simple *2 gives me completely false results. I've attached a simple example patch to show what I am trying to achieve. I used the mean object for the simplicity of the example, but I would need all the numbers coming out of pipo~ (I use them to control some markov chains). 

I'm not sure pipo~ is really meant for such precision, maybe there are other objects more suitable for this?

Hi everyone,

I'm working on a simple AI, analyzing example sounds to then try to reproduce them. My first step is to analyse the audio exam

getting-consistent-audio-descriptors-while-playing-sounds-faster

Have you looked at the Flucoma externals?

For most you can analyse a buffer instead of real-time, so perhaps a solution for you...

Have you looked at the Flucoma externals? 

Thanks for the answer. I don't know why I didn't thought of analysing buffers instead of real-time... I guess it was too easy. 

Anyway I've seen Flucoma pass by a few times now, I'll have to check it out!

Hi Mark,

Thanks for the answer. I don't know why I didn't thought of analysing buffers instead of real-time... I guess it was too easy. 

Anyway I've seen Flucoma pass by a few times now, I'll have to check it out!

 is made for that. You load the files to be analysed in a mubu container (don't bother with buffer~, as it can only load one file) and analyse all of them as fast as your CPUs can.

Hi Jean-Philippe, mubu.process is made for that. You load the files to be analysed in a mubu container (don't bother with buffer~, as it can only load one file) and analyse all of them as fast as your CPUs can.

Getting consistent audio descriptors while playing sounds faster