It's common in modern battery-powered audio electronics not to reference audio to 'true' ground, with +/- supply rails. Instead, you make an audio reference half way between ground and your voltage rail (often with a pair of resistors and a buffer of some kind). Then you arrange your audio signal to swing either side of this mid-rail reference - one tends to end up with quite a lot of series coupling capacitors to let you shift the DC point of the signal around.
Alternatively, if you don't like that (and there are reasons not to - getting the reference quiet enough and stiff enough can be a pain, as can the series caps, and at 3v3 you're fairly short of headroom once you've cut it in half), then a -ve rail can be produced using a charge-pump device like a MAX660 (there are lots of other sources for basically the same part) - that's a lot simpler than a full-up DC-DC converter.