How to detect Musical keys such as (c,d,e,f,c# etc) from audio using librosa ?

[Prashant Kumar]

Is there any class to detect keys in a audio file ?

[Brian McFee]

librosa does not currently have a key detection model.

However, it ought to be easy to implement a simple model using the pre-computed features at https://www.ee.columbia.edu/~dpwe/e4896/practicals.html#prac10 along with the key labels available at http://isophonics.org/content/reference-annotations-beatles .

[Rafael Valle]

Brian, 

while at labrosa last summer I implemented key detection algorithms, including these ones in the literature:

1. Krumhansl, C. "Cognitive Foundations of Musical Pitch, ch. 4" (and Schmucker)
   
2. Madsen, S. T. and Widmer, G. "Key-Finding With Interval Profiles"
   
3. Temperley, David. "A Bayesian key-finding model."

I've been planning to do a pull-request but wanted to check with you first if this is of interest to librosa. 

Please let me know!

[Brian McFee]

Rafael,

That sounds great. Maybe start an issue on github and we can discuss in more detail?

Generally speaking, I'd like to avoid adding functionality that has strong statistical dependencies on specific data sets. If any of those key detection models so require fitting to specific data, they'd be better done as addon/contributed modules.