0. Preamble

AUSS is meant to be designed with the end developer in mind. The target market is people using opensource audio software who want to link various opensource components together. I am not restricted AUSS to only opensource but it is the "market" which is the most relevant.

AUSS is easy to develop for because at the very least all one needs to do is using pipes to stream audio between applications, you don't even "need" a header. As long as the user is aware of the types of audio streams they are using the system will work easily. Potentially one could just print audio bytes to stdout which could be piped into a sound stream player. Your audio printing program would then be playing audio is realtime!

AUSS is multi-language. You can use any language which gives any of these features; writing to stdout; reading from stdin; pipes; sockets; tcp connections,... You can use languages which the library hasn't been ported to just because you can print!

Sockets are used in many cases instead of pipes because:

• Sockets are bidirectional so less links to do more work.
• Sockets are available on Unix systems as well as Windows (which we
• hope to port AUSS to later).
• Pipes have HORRIBLE performance on Windows (again for future porting).
• Sockets using TCP/IP can be used over a network.

Each Frame consists of 1 sample per channel of audio. 1 Sample is the size of the number representing the amplitude of the audio at one point in time. A frame of CD AUDIO quality sound consists of 2 16bit samples, for Left and Right channel. What AUSS enables users to do is to pass raw frames between processes easily, because you can do this your program doesn't need to interface.

Optionally headers can be included in front of an audio stream (or in the opposite direction of a stream accepting audio). These headers are in XML but MUST be padded to be a the size of a multiple of frames. This is so that systems which don't support AUSS headers can still play the sound stream, this means the headers will be played as if they were normal audio. At most this will be a little pop. By being the same size as a multiple it makes sure all frames are properly aligned and interleaved data does not lose sync. If the header wasn't a multiple potentially you could have even the samples not being aligned so would hear noise rather than the sample itself.

Headers are useful for AUSS programs which would like named connections, programs like the Connector, otherwise most AUSS programs will just want anonymous connections.

Most of the header is optional. The main point of a header is pass an audio stream's name along.

There are 3 ways to talk to the connector:

• Input Connection: Provide a sound stream to the connector.
• Output Connection: Receive a sound stream from the connector.
• Filter Connection: Receive and provide a sound stream to the connector. A Filter connection does not necessarily have to act as a filter.

In the case of the Output connection a header maybe provided if underlying link is by directional (a socket). The Connector will check for a header, once the header is read in the connector will ignore any further bytes.

Once a program is connected to the Connector, a program can read and write audio depending on the connection type. If there are no bytes to read the program should block until there are bytes since the system maybe outputting to the soundcard and only so much data can be passed to the soundcard at once.