MATLAB implementation of the model that you're referring to can be found here:
http://www.petitcolas.net/fabien/software/mpeg/
And I think it should be easy to modify the tonality detection like the way you want it to be.
Regards.
Arijit
**Disclaimer** The sender of this email is NOT authorized to represent or act on behalf of Nanyang Technological University in any matter or transaction by means of using this email account. Nanyang Technological University will not be responsible for any content or views expressed in this email and this email does not express the views or opinions of the Nanyang Technological University.
From: AUDITORY - Research in Auditory Perception <AUDITORY@xxxxxxxxxxxxxxx> on behalf of Goran Markovic <skromnibog@xxxxxxxxx>
Sent: Thursday, October 23, 2014 11:35 AM
To: AUDITORY@xxxxxxxxxxxxxxx
Subject: Tonal component detectionDear auditory list members,
in the psychoacoustic model 1 of ISO/IEC 11172-3 (the one used for MPEG Layer I/II) a tonal component is detected as being local maximum and for at least 7 dB above it's largest neighbour.Do you know of any publicly described tonal component detection where a tonal component is detected by comparing a local maximum with the average of the neighbours (i.e. comparing it to the spectral envelope)?
Best regards,Goran