There are definately LEDs that operate at different wavelengths, but a detector will have peak sensitivity at some wavelength (which hopefully is the wavelength it was designed for) and then the sensitivity will dropoff from the peak. So if you have a 422 and 633 next to each other along with their respective detectors, they will both see each other, but at different levels. There are optical filters, but you might be able to do this another way.
Perhaps you should provide your desired result. In the meantime, let me offer this idea:
Using two LEDs (which could be identical), you modulate the light to each one by two different frequencies. Now you can use a single detector coupled with a pair of bandstop filters to pull out the signals at each of the two frequencies. Alternatively, if the frequencies are low enough, you can sample them and perform an FFT in software. Note that since you are only creating 2 bands, you could perform the discrete DFT on just the two frequencies of interest and possibly save some cpu cycles.