If you have a particular sound that is embedded into a wave file containing other sounds, the only mechanism anyone has for isolating that sound and excluding all other sounds is through filtering BUT this probably won't work in a general sense because....
A particular sound may be (say) a flute and its characteristic timbre is produced by a series of harmonics related to the base pitch of the note - these harmonics will extend through the audio spectrum and overlap other instruments that are also in the wave file. Disentangling a flute harmonics from other sounds is very, very hard to do effectively.
We, as humans, may recognize that a flute is being played (in the presence of other sounds) and if we can play the flute we could recreate that sound fairly well BUT it would be an interpretation of the flute part of the wave file.
How do you instruct a piece of software or hardware to recognize a flute sound and extract it - I don't think it can be done by filtering because the harmonics would clash with other instruments.
If the sound you wish to isolate is limited to a certain range of the spectrum and other sounds in the wave file do not encroach into that part of the spectrum then simple band-pass filtering should be successful. Anything more complex and the process is rather like trying to turn a baked-cake back into it's original constituents.