Now that you've got some ideas of hardware from here, you'll want to go back to the customer and have a more detailed conversation about expectations. If they're thinking that there's a magic box out there that can take any sort of audio input they
throw at it and turn it into full on rock concert lighting, then the answer is "no". There's a reason lighting designers and programmers still have jobs.
ETC Mosaic is the product I'm most familiar with for this kind of application (but as mentioned, it's basically the same as Pharos). It's pretty flexible, and can
monitor the audio input across several configurable frequency bands, then have things happen based on the
level in each frequency band. The simplest way of looking at it is if you have one light for each frequency band, then the
level could control the brightness of each
fixture. Or it could be hue rather than brightness in a color-changing
fixture, or it could be several fixtures forming a sort of bar-graph. You can also set triggers to do more complex actions/effects when a
level is exceeded (ie: do something flashy on each big bassbeat--and therein lies one of the big problems: it's not reasonable to try to pick thresholds that would work well for every kind of music played at any
level. Even if you're doing a really basic level-to-brightness
effect, you still have to pick a scale. If you adjust for louder songs, then quiet songs will never result in useful brightness. If you adjust for quiet songs, then loud songs will result in everything being stuck at full. Similarly, a lot of modern recordings are so compressed that there isn't enough
dynamic range to do anything interesting at all regardless of what settings you choose.
If you have a limited selection of music, can afford to spend a lot of time tweaking settings, and can count on it always being played back at the same
level, then you've got a chance of doing something at least sort of interesting. If not all of those, then it's going to be pretty mediocre.