Several companies have tried this. They were crushed by ever falling prices of polysilicon.
The easy way to do this is to get a fluorescent material. Fluorescence occurs when photons from a wide range of frequency are absorbed and then re-emitted at one strongly favoured frequency. However, this process is generally a "downconversion*, the output is a lower frequency than the input.
This is still very useful. For instance, most laundry detergents include a fluorescent dye, which absorbs UV light and re-emits it as blue light. This is why they're almost always blue. The blue light offsets the yellowish tint that old clothes get, and yellow + blue = white, so now your shirt looks white again.
In the case of solar panels, this only helps if your cells are really efficient at a particular frequency. In the case of silicon we're in luck, because that's true - silicon cells would work better if all the sunlight was red instead of a broad spectrum. Actually it's true for most solar cells, they will always be most efficient for light at a specific frequency.
Now this was a big idea when silicon was expensive. One idea, the "Slivr", used sheets of dyed plexiglass that glowed in red. They were arranged so light hit the plastic at an angle, so that it became trapped within the plastic (total internal reflection). As it bounced around it would eventually hit the dye and downconvert into red. There was a small number of cells on one end, so these small strip of cells got light from the entire plastic sheet.
Now this process is NOT that efficient. Some of the light goes right through the sheet, some reflects, a bunch is absorbed. But with the downconversion you're getting some of that back. And at the time, silicon was way expensive, so even though you might not be getting all the light you would if it was just a cell in direct sunlight, you're still getting a whole lot of light for the cost of a small cell.
The only thing is, it turns out you can make pSi for not much more than a sheet of dyed plexi! And so Slivr is no more. There was one in the US as well, same basic concept, but I don't recall their name.
Now the opposite process also occurs, parametric upconversion. This is much more difficult to arrange, and the cost of the things that do it are way out there. So no one does that, that I know of, for solar cells. They do use them in lasers though, because it's easy to make a powerful IR laser and then upconvert that to what you're looking for. The National Ignition Facility uses this to turn it's IR into UV, for instance.
UPDATE: Nope, SLIVR is another design, I have the names mixed up.
