In composition, there is a fine line between "finding your voice" and painting by number; between giving each piece a personal touch and writing the same piece over and over. As a composer, I have never been particularly interested in larger guiding principles such as sonata form, serialism, metric modulation, etc. even though there are plenty of pieces I enjoy by other composers which use these devices. In fact, I usually enjoy listening to them right up until the point where I start to really hear the method, at which point the music always loses a little bit of its luster, at least to me.
Of course, some if not all of these devices were codified as such after the fact, but even so, later composers often pick them up and embrace them as part and parcel of their compositional technique. Not to say that this is wrong in any intrinsic sense, but I have never understood this, nor have I ever felt compelled to do it, even when the method is of my own invention, as in the piece I wrote which followed a nearly strict pattern of meter changes: 7/4 5/4 3/4 -or- 7/4 5/4 7/4 5/4 3/4. This led to a cool "counting down" effect whereby the bars got progressively shorter; the process would then repeat itself. Among all of my compositions, this idea had the most possibilities for further development; however, I have yet to find what I deem a legitimate need or even a suitable use for it outside of the piece from which it arose as an intuitive creation. That's just how I am.
I believe that the "personal voice" operates on deeper level than the compositional device. I tremendously enjoy approaching each subsequent piece as an opportunity to create something entirely different from everything else I have written, and it particularly intrigues me when composers say things along the lines of "Oh, I've moved on to x approach" or "I only work in x medium/style now." I always want to ask, "What was it about the previous approach was preventing you from taking the new approach?" I'd like to think that the ideal for a composer would be to be able to answer "Nothing," meaning of course that you would not need to change approaches midstream in the first place.