The HTML5 standard is widespread on mobile devices. In combination with the Web Audio API, it allows for massively distributed real-time audio rendering. But timing issues exist, mainly because of the lack of standard inter-device synchronisation. This paper proposes a synchronisation solution based on HTML5. Using a shared reference time, we achieved the distributed rendering of audio events with an individual accuracy of 1 to 10 ms, 5 ms in standard deviation, which is more accurate than the audio block duration, for any device that we measured.