Topics

Ethnomusicology/Music Analysis

Alex Enkerli <aenkerli@...>
 

Hi,
I know this list mostly deals with technical questions but are there other
people here who use Praat to analyze music?
Thanks.

Alex Enkerli
Doctoral Candidate
Department of Folklore and Ethnomusicology
Indiana University

firegold@...
 

Can't help you with your question, but got one of my own. How are
you using Praat to analyze music, and what sorts of things are you
looking for?

Just curious.

--- In praat-users@y..., Alex Enkerli <aenkerli@i...> wrote:
Hi,
I know this list mostly deals with technical questions but are
there other
people here who use Praat to analyze music?
Thanks.

Alex Enkerli
Doctoral Candidate
Department of Folklore and Ethnomusicology
Indiana University

Alex Enkerli <aenkerli@...>
 

firegold@...:
How are you using Praat to analyze music, and what sorts of things are you
looking for?
Well, as of yet, I've mostly used Praat for segmentation and labeling of
songs that I have. I can then use the temporal information for different
things (measuring tempo, describing structural events...). The way I do it
can be as crude as tapping boundaries while playing the signal.
Further things I'd like to do might include automatic labeling, beat
detection (someone on the list sent me a reply about that), filtering, and
standard acoustic analysis (spectrographs, pitch-tracking, etc.). I'm still
in an exploratory mode as I'm preparing for a field trip.
What I'm studying are praise-songs performed by hunter-bards in the Mande
region of West Africa.

So I was wondering if other people did music analysis and what kind of work
they might do. Are you involved in anything musical?

Alex Enkerli
Doctoral Candidate
Department of Folklore and Ethnomusicology
Indiana University

Bernard Bel <bel@...>
 

Alex Enkerli, aenkerli@... wrote:

So I was wondering if other people did music analysis and what kind of work
they might do.
I am currently doing the analysis of songs by women at the grindmill in Maharashtra (India). It is a comparative study between melodic patterns of "semi-improvized" singing, and melodic excursions in spontaneous speech -- however, the latter is from a French corpus. We are trying to outline strategies for conveying emotion... We have several articles in press. ("We" includes my colleague Genevieve Caelen, who has been in the field of speech prosody for 30 years.) You'll get an idea of vthis field visiting <http://www.ccrss.ws>.

My initial interest in PRAAT was its efficient pitch-tracking algorithm. So far it is the only program I have tried that successfully extracted the pitchlines from our recordings. We record songs with fair microphones and a DAT, but the background is quite noisy. Notably we pick up a lot of sound from the hand-driven grindmill itself. We can't ask women not to grind because we want "real" conditions, and we know there work gestures are related with the time structures of singing. Nonetheless we also have recordings without grinding.

So, I was delighted when I saw melodic lines displayed on the screen after very little fiddling with parameters. A few years earlier I had spent weeks trying to program a pitch extractor, but the results were poor. Before that, the best pitch extractor I ever used for singing voices was a piece of hardware I had built in 1982, the "Melodic Movement Analyzer". It did the job in real time and sent its data to an Apple II for storage and analysis. But, well, do you still have an Apple II ?

Sorry for being nostalgic about glorious years! Congratulations to Paul for helping us to revive our analytical work without electronic junk...

At this early stage, our work on the songs is still descriptive, but we will gradually take advantage of many other features of PRAAT when it comes to quantitative analysis.

Bernard Bel

Bernard Bel <bel@...>
Laboratoire Parole et Langage
ESA 6057 CNRS - Université de Provence
29 av. R. Schuman
13100 Aix-en-Provence (France)
------------------------------
Visit LPL: <http://www.lpl.univ-aix.fr>
-----
Compose Music on the Mac! <http://www.bp2.org>