Audio streaming server which encodes from sound input

I would like a solution that runs on Linux (preferably Debian, but any mainstream distro) which encodes audio coming from up to three audio inputs plugged into the server, and streams that in audio formats which are compatible for playback on an iPhone, iPad or Android client. The clients will be connected via a WLAN to the server, and would browse to a simple mobile-compatible web page on the local network. The web page would have the three links to the streaming audio which is being generated by the server. All components of the software stack must be open source/ free software. The final application developed by this project will be owned by me, however. You must also specify the hardware which can run this solution, and it must be sourced in Australia. Particularly important is the specification of an audio card/s which can input three audio signals so the software solution can encode and stream three at once. The hardware has to be compatible with the specified Linux distribution too. It would be nice, but not mandatory, for the hardware to be quiet (eg. slow fan speed) and generate minimal heat relative to big servers. Hardware specification can be a currently available brand-name computer by model name, or a set of hardware components which need to be put together but are known to work together (eg. motherboard, power supply, RAM, CPU and case specs). It should be low cost, should have at least a year's warranty on the hardware from whichever manufacturer provides it. It should have enough performance to do just the encoding job and webserver job. There will be up to 20 users listening to the streaming audio, so the solution should be designed with that capacity. You must provide a set of batch/configuration scripts which installs all the software. If ffserver, ffmpeg, and any o/s patches are required, the script should include commands to do all installation and updates necessary. Ideally one command line instruction is required to initiate the entire installation process, and a success or error result is returned. You must also provide the simple web page (this just needs to be functional, no graphic design required) which includes links to the three audio streams. The page must be compatible with iPhone and Android phones, iPad and Android tablets. The server should be configured to serve that web page with Apache or similar. I expect that page would be served on the local network (eg. [login to view URL]) to mobile clients who have joined the local WLAN network. The audio does not need to stream out to the internet, it is only streaming to local users. Once a user clicks on one of the three audio streams on that web page, the native player on the mobile device can take over so the user can hear the audio. The application should automatically activate if the server is restarted after a power failure. The software solution can be developed in any mainstream language. I expect there is minimal development, the job I imagine is mostly configuration of ffserver into a batch script, activating it as a service. I would prefer ffserver and ffmpeg to be used because it's Open Source and it appears to be the correct software application for this project, however if a better alternative exists then make a recommendation [login to view URL] The server will have internet access.