The -N=N- group has bands around 1400cm-1. However depending on the molecule and pH it can switch to the hydrazo tautomer and in a dye with two azo groups you can have one in the azo form and the other in the hydrazo form which affects the band position. I have attached two papers on the SERRS spectra of azo dyes which you may find useful

Raman spectra of azo dyes can be theoretically modeled in software, e.g. Gaussian. These can sometimes be enough to give a good approximation of the vibrational modes in SERS spectra of dyes.

https://www.researchgate.net/publication/231396277_Vibrational_Analysis_of_trans-Azobenzene

https://www.researchgate.net/profile/Siva_Umapathy/publication/231666036_Structures_Vibrational_Frequencies_and_Normal_Modes_of_Substituted_Azo_Dyes_Infrared_Raman_and_Density_Functional_Calculations/links/02bfe511a0cee4dfdc000000.pdf

SERS spectra are more complex to model as you need to include a silver cluster as well as the dye - see example below:

http://dspace.xmu.edu.cn/bitstream/handle/2288/62936/A%20DFT%20study%20on%20photoinduced%20surface%20catalytic%20coupling%20reactions%20on%20nanostructured%20silver%20selective%20formation%20of%20azobenzene%20derivatives%20from%20para-substituted%20nitrobenzene%20and%20aniline.pdf?sequence=1

Article Vibrational Analysis of trans-Azobenzene

Article Structures, Vibrational Frequencies, and Normal Modes of Sub...

Thank you very much.