I wouldn't multiplex the voice into the data fiber for much of the reason ScottMac points out. Such equipment is phenomenally expensive once you multiplex ATM with SS7 and other data/video.
It's better to use separate pairs of fiber dedicated for the voice and separate from the data. Yeah, there would be a lot of unused bandwidth on the fiber, but it's easier to setup individual voice to voice multiplexors with fiber converters than to multiplex multiple T1's into the same fiber pair used for data.
Personally, I've never used any Paradyne multiplexing equipment, but it exists. This would be data/voice/video into fiber and vice versa. I've used and installed some of their other equipment and their equipment is good.
RAD makes a good assortment of multiplexers for data/voice/video. If you have T1's or multiple T1's to multiplex, I recommend their solutions. With this example you may also need RADview (a NMS) or SNMP. Here's an example of what I mean:
http://www.rad.com/products/family/opt_4t1/opt_4t1.htm Verify something like this with their consultants (salesmen). Yes, they know the capabilities of their equipment. If you have more than 4 T1's from point to point, their Optimux-T3 may be what you need, just don't try to add data.
BlackBox also makes converters and multiplexers of various types, but not in the T1 range.
--
Oh, btw, check into whether you have channelized T1's/T3's or otherwise for you voice drop point of entry or demark. And then talk with consultants accordingly.
---
If you use one pair of fiber for the ATM network (if the campus has one) and another pair is reserved for voice, this product by RAD allows you to mix 1-10Mbit and 1-10/100 Ethernet with up to 8 T1's or other combinations like HSSI. (Modular)
http://www.rad.com/products/family/opt-xlt1/opt-xlt1.htm Oh, btw, on this equipment and the RAD equipment above, they have single fiber capability (using two wavelengths on each fiber). This is the SF1 and SF2 combinations of equipment. I first used such configurations in 98 and they were nifty then in saving fibers.
Nothing is cheap, but this is a starting point for research and comparison. Data has significantly higher bandwidth requirements than voice. If you are doing this for school from elementary through high school, my guess is that you need a relatively low voice capacity with standard data capability.