For food: 3-4 kiloWatthours per day per person. The energy requirement for each crew-member is about that of 2-4 desktop PCs.
For other objects: Far more power than the replicator was designed for.
As laid out by @Horuskol, we assume that the food is somehow assembled from simpler components. If say, carbon, hydrogen and oxygen are assembled into fat molecules, these molecules contain more energy than the original atoms and the replicator must have added that energy. This is equivalent to the "food energy" released when a human consumes it.
A human gets about 12 MegaJoules (MJ) worth of energy from her food every day. This is equivalent to ca 3.3 kWh or 2 desktop PCs running all day.
We don't know the energy requirements of teleporting the food into the replicator tray, but since we're not seeing much dissipation of energy at the receiving end (heat, radiation) it's fair to assume not much energy was expended at the sender either. This leaves the food energy as the main energy cost.
What we have not calculated is the energy cost of liberating individual atoms from the feedstock. For food, the required Hydrogen, Oxygen and carbon can be stored as gases (H, O2 and CO2). But if you are going to replicate something made of metal, your energy requirement is basically that of vaporizing the metal.
To vaporize 10kg of iron takes about 635 MJ or more than 50 days worth of replicator food. Assembling this in a second, as replicators do, means applying 0.6 GigaWatts of power. In today's terms, it is equivalent to the power from 300 windmills or roughly half a nuclear plant.
Assuming 4 meals a day, assembling something with 10kg of iron therefore taxes the replicator 200 times as much as assembling a single meal. This explains why food is easily replicated, but heavy gear and tools are out of the question.