Method for comparing sounds in max/msp

Fever's icon

Hi.

I'm having these two sound files and now I'm looking for some different methods for comparing em', to find out how similar they are.
Anyone know if there are any easy methods available for that kind of operation in max/msp?

Charles Baker's icon

M. Christensen: RE: comparing sounds

hmm...this is not as simple as it seems:
Similar in what ways?
transient analysis/ beat tracking would give clues as to tempo,
rhythmic / arhythmic, metric analysis:
but this is not a simple process.
Pitch tracking might reveal issues with pitch/harmony/keys,
but if the sounds are polyphonic this is *very* difficult.

I think you need to ask the question my reply started with ...
similar in what ways?
and when you can clearly state that, go looking for analysis tools
that address this.
There was a recent Computer Music Journal issue (v28:3 Fall2004 )
that addressed
issues in this area as part of the even larger area known as "Music
Information Retreival":
What would a query to find a sound :"like" another sound have to have?
There are many approaches, and they all differ a little in how they
answer the question:
Similar in what ways?

just my 2cents,
j2k

robotic-audio's icon

some ideas in the thread called Scrambled HackZ ideas, not long ago...

Some analyse externs

Beat~
Bonk~
Analysis~
Fiddle~
Pitch~

and muck more...

search www.maxobjects.com

Fever's icon

Ohh I see. Never really thought of similarity in such a complex way. The similarity I look for, would prolly just be in something like frequenzy, amplitude or pitch.

robotic-audio's icon

then you should check out

fiddle~ & pitch~

Fever's icon

I will try that. Thank you. :)

Stefan Tiedje's icon

Morten Fever Christensen wrote:
> I'm having these two sound files and now I'm looking for some
> different methods for comparing em', to find out how similar they
> are. Anyone know if there are any easy methods available for that
> kind of operation in max/msp?

I'll recommend to read "Goedel, Escher, Bach" from Douglas R.
Hofstadter, which shows how difficult it is to compare fonts, which
seems easy for humans. To compare sounds seems even difficult for humans...

But there is hope, we had the thread about the scrambled hackz idea...,
which needs some kind of comparison between sounds. You'd need to
analyze the sound in different ways, maybe teach a neural net...

Anyway let us know with what you come up. It would make up a great
postdoctoral thesis...

Stefan

--

[][] [][][] [][] [][][]
[][][][][][][][][][][][][][][]

Stefan Tiedje
Klanggestalter
Electronic Composition
&
Improvisation

/~~~~~
\ /|() ()|
))))) )| | |( \
/// _/)/ )))))
___/ ///

-------------------------x----
--_____-----------|-----------
--(_|_ ----|-----|-----()----
-- _|_)----|-----()-----------
----------()------------x-----

14, Av. Pr. Franklin Roosevelt,
94320 Thiais, France
Phone at CCMIX +33-1-57 42 91 09