Midi, as The Chas says, is a set of instructions that tell a MIDI enabled instrument, or midi synth on the soundcard, to play notes in a given sequence. WAVS are actual waveforms that are converted to sound by your speakers. With midi files, there are no waveforms in the file...these are provided by the soundcard or instrument.
A typical audio signal has a frequency range of 20Hz to 20KHz, and to 'sample' this range of frequencies, the sample rate must be at least double the highest frequency. A typical sample rate for audio is 44.1kHz or so. The level of the audio waveform is stored as a binary number, the number of bits giving the number of discreet values that can be represented. 8 bit resolution gives 256 levels, and 16 bit resolution gives 65536 levels. The higher the resolution, the better the accuracy or quantisation of the original sound. So, a stereo source, sampled at 44.1k with 16 bit resolution needs 44100 x 2 x 2 bytes per second of audio in its rawest form, or approximately 176 kilobytes. MP3 and other forms of compression use algorythms to reduce the file size, but the greater the compression, the less detail can be stored.
With midi, the instrument sound samples are stored in ROM and the midi file just tellsthe instrument which samples to play at a given time...the samples are not embedded into the file. Hence MIDI files are just a few Kb in size. This also explains why you don't get any vocals in MIDI!!!!