A theory has been developed to show that the FM component can be extracted using second-order all pole modelling, and a technique for extracting FM features using this model is proposed, to produce very smooth, slowly varying FM features that are effective for speaker recognition. This is evaluated on NIST 2008 international speaker recognition evaluation. Optimum Frequency band allocation specifically to capture speaker specific information is studied in terms of the number of sub-bands and spacing of center frequencies, and two new frequency band re-allocations are proposed for FM based speaker recognition. Furthermore, FM feature is successfully utilized for automatic forensic speaker recognition, which is implemented based on the likelihood ratio framework with two stage modelling and calibration, and shown to behave in a complementary manner to amplitude information.