Audio over Internet using OSC

Audio signals should be distributed to playback-systems over network. To save resources like bandwidth, and computing time the audio data should be sent on demand only.

Ethernet established itself quite well in the field of controlling embedded devices and computer. Furthermore most of the available network-components are able to handle the IP-protocol. Open Sound Control (OSC) will be used as transmission protocol, due to its message based structure. Unfortunately there is no standardized implementation for audio via OSC yet. So with regard to the possibility of synchronicity and partly discontinuous data stream with constant and predictable latency a protocol to transmit audio over OSC has to be developed. As remote terminal microcomputers, embedded devices (e.g. microcontrollers) and computer-workstations with integrated ethernet interface are used.

A microcontroller-board, able to receive a variable amount of audio channels over ethernet, should be implemented. The digital audio data will be converted into the analog domain via pulse-width-modulators (eventually linearized), which control FET-output-stages. The analog audio stream should be played back by integrated amplifiers, optionally implemented as Class-D amplifier. Generally the whole board should be a low-cost device.

The device should respond to TCP and UDP and on top of this DHCP and NTP should be additionally supported. The audio data is organized in OSC-Packets and those packets are again embedded in a TCP or UDP container. Packed in bundles with a time stamp the data could be sent to our target directly via unicast or parallel using multicast over ethernet.

The development of the data-protocol will be tested in a network environment with audio computers. Subsequently the implementation on the microcontrollerboard follows.