This info was copied from the M-Audio site and unless you're good with a soldering iron and comfortable with electronics, just return your UNO to where you bought it and get something else.
Thus, continue at your own risk (resistors can be 1/2W or 1/4W (these are smaller):
The eight pin chip on the UNO's circuit board labeled U2 is the optocoupler that takes current pulses from your MIDI device and turns them into voltage pulses that are decoded by U1 and sent to the computer via USB. The optocoupler in the defective UNOs is a 6N138. The problem is one of two things (depending on M-Audio's original design). Either, the factory installed the wrong optocoupler (which is likely because the 6N138 is not a very good choice for a MIDI input), or M-Audio forgot to connect pin 7 of the optocoupler to ground with a resistor (which means the optocoupler's output turns on fast, but takes way too long to turn off, effectively causing it to miss pulses).
So, you've got a defective UNO with the wrong opto and it's missing a resistor (but you've got a soldering iron and a do-it-yourself attitude), what's the solution? If you solder a 5k ohm resistor from pin 7 to pin 5 on the optocoupler (6N138, U2), your MIDI input should spring to life. Adding this resistor provides a discharge path for the parasitic capacitance on the base of the output transistor in the darlington pair. Discharging the base through a resistor gets the output transistor turned off fast (no missed pulses). I added a 5k ohm resistor to mine and now it works great. Like I mentioned, the 6N138 is not a good choice for MIDI. It is a slow device with a high CTR, while MIDI (5mA input, 31250 baud, low output current) really screams for a fast device with a low CTR, but I used an oscilloscope to verify the output of the opto with a 5k ohm base resistor and it looks great.
Extra credit: While troubleshooting the MIDI input I also noticed that the leakage current of my keyboard's MIDI output was still enough to keep the input diode of the optocoupler turned on (slightly). It was not on enough to cause a problem on the output, but it sure wasn't off. This problem can be solved by soldering a 10k ohm resistor in parallel with the input diode (i.e. from pin 2 to pin 3 of U2). This resistor provides a path around the input diode for leakage currents up to 100 microamps (without developing the 1V forward voltage required to turn on the input diode).
Solder a (4.7K is close enough) 5k ohm resistor from pin 7 to pin 5 on chip U2. Optionally, solder a 10k resistor from pin 2 to pin 3 on U2.
1. crack open the case with a fine tip screwdriver
2. identify the correct pins on the U2 chip. In the photo of the top of the board, you can see the black dot on the U2 chip; that is beside pin #1 and then the pins number counterclockwise around the chip (viewed from above).
3. solder the resistors on. Resistors are reversible so it doesn't matter which way they are oriented. I soldered mine to the back of the board because there was more room there. I also trimmed off about half of the leads on the resistors to make them a little smaller.