A very interesting project that I need to check out:
SoundCrusher Audio File Manipulation Framework by Marek Bereza.
To quote the above web page, SoundCrusher is:
an object-oriented framework that provides manipulation operations on audio files in many different formats. It builds another layer of abstraction on top of libsndfile, which is an audio file i/o library, so SoundCrusher supports all the formats supported by libsndfile.
It sounds like the idea is to facilitate fast editing of large sound files - eg, I’ve often had the problem of top’n'tailing say one hour of audio - ie snipping a section from the start and end and saving the result to a new file. The save always takes a long time as a temporary copy of the original file is altered in virtual memory and saved out to a new file. It sounds like SoundCrusher is able to provide a quicker save by performing the edit just by rearranging a list of pointers to regions within the original file.
I have yet to try it out, but will post back my findings when I do.