Don't look for a specific code (program). They should generally perform the same under the same circumstances. I mean if you need help in assigning the observed (FTIR) peaks, first choose a reliable method, i. e. either DFT (B3LYP or B3PW91) or HF or a semiempirical..., try to simulate the spectrum by computing vibrational frequencies, scale the (low, high or all) frequencies using appropriate scaling factors (there are papers and tables, as those maintained by NIST, http://cccbdb.nist.gov/vibscalejust.asp). Finally, decompose the modes using a PED (potential energy distribution) code and look for a match. There should be one, although computed frequencies will not always fit in few cm-1.
GAMESS is free and as I know well suited for such calculations (may be a little difficult to work with as a beginer). You may search also for a graphical interface program to use it with GAMESS (such Avogadro, Gabedit, Dragon..etc)....I would prefer Gaussian (with GaussView as interface) but it doesn't come with free licence. ACD Spectrus seems very interesting but yet is not free as I suppose. Also I think that ChemOffice package has IR and NMR prediction but is not free also (shareware), however it is OK for some predictions..(for NMR I wouldn't use it for complicated molecules).
Here you can find spectral resources of known compounds (including IR)
hi , I'm wondering why u need to use virtual IR program . as u know the chem draw or hyper chem programs don't give u a good agreement with it's actual values . beside to a lot of factors connected with stretching frequency for each group.so I suggest to use actual FT-IR spectra with it's accesses. good luck
ACD Spectrus processor can help you in suggesting you the IR frequencies of a functional groups in a molecule http://www.acdlabs.com/products/adh/spectrusprocessor/optical/#.
Don't look for a specific code (program). They should generally perform the same under the same circumstances. I mean if you need help in assigning the observed (FTIR) peaks, first choose a reliable method, i. e. either DFT (B3LYP or B3PW91) or HF or a semiempirical..., try to simulate the spectrum by computing vibrational frequencies, scale the (low, high or all) frequencies using appropriate scaling factors (there are papers and tables, as those maintained by NIST, http://cccbdb.nist.gov/vibscalejust.asp). Finally, decompose the modes using a PED (potential energy distribution) code and look for a match. There should be one, although computed frequencies will not always fit in few cm-1.