[LAU] Analysis of monophonic audio signals on the commandline

[Fernando Lopez-Lezcano]

Hard to add anything other than what Fons, Bill (his birds.clm are 
great!) and others have written.

I thought I should maybe mention ATS (Analysis, Transformation, 
Synthesis) and associated programs:

https://dxarts.washington.edu/wiki/analysis-transformation-and-synthesis-ats

There is a command line analysis tool that can generate the analysis 
files (a mix of sinusoidal components and bandlimited noise bands). I 
have used it successfully for some of my music in the context of CLM and 
SuperCollider (there are SC UGens that can read those analysis files).

In any case it shares all the limitations of analysis that have been 
mentioned before. Maybe slowing down the bird songs and then analyzing 
them, as Fons suggested?

Can't wait to hear the results!
Best,
-- Fernando


On 2/27/22 5:48 AM, Fons Adriaensen wrote:
> On Sun, Feb 27, 2022 at 02:06:06PM +0100, Jeanette C. wrote:
> 
>> At least, analysing the song of the chaffinch is free of broadband features.
>> Certainly going by ear alone. That together with some more already coded
>> data from other bird calls may greatly help in a semi-analytical approach,
>> which may yield good enough results.
> 
> For us humans, the gray area between time-domain and frequency domain
> features is around 20 Hz. For example a 10 Hz modulation we will hear
> as vibrato, but 30 Hz certainly will give a very different impression,
> that of a rough sound or a complex spectrum.
> 
> Since birds are much smaller than humans, they can produce faster
> modulations and they also tend to generate higher frequencies. Very
> probably for them the gray zone occurs at a higher frequency.
> 
> That means that of the most revealing ways to analyse bird sounds
> is to listen to them slowed down (not time-stretched). That will
> allow to hear modulations that are too fast for us to be perceived
> as such otherwise.
> 
> Ciao,
>