This looks very promising! They claim to psycho-acoustically outperform AAC and Vorbis, at much lower latencies, with good packet loss compensation. And BSD-licensed reference software. This could really boost the practicality of realtime audio streaming over a wifi LAN.