Inferno-AoIP ALSA PCM module (virtual soundcard)

1. Follow common instructions in Quick start to build alsa_pcm_inferno and start clock synchronization daemon.
2. Copy (or symlink) libasound_module_pcm_inferno.so from target/debug/ or target/release/ directory to /usr/lib64/alsa-lib/ (Fedora) or /usr/lib/x86_64-linux-gnu/alsa-lib/ (Debian, Ubuntu). If using CPU architecture other than x86_64, the target directory will be different. find /usr/lib* -type d -name alsa-lib if unsure.
3. Add pcm. device with type inferno to your .asoundrc (example asoundrc)
4. In the application there should be a place where usually device name is entered, e.g. hw:1. Enter the name of the device created in your .asoundrc there. If you've copied the example asoundrc, the name is just inferno.

Application must support 32-bit signed integer audio samples. (Audacity doesn't, but generally using Audacity directly with this plugin is not a good idea) plug plugin shipped with ALSA should work as automatic converter for apps that don't support that format (not tested yet).

To change the number of channels, specify the environment variables INFERNO_RX_CHANNELS and INFERNO_TX_CHANNELS.

If you don't want to change anything outside your $HOME, you can set ALSA environment variables to make libasound search for configuration file and modules in custom directories. Example is in test_effect_processor.sh. This will replace the whole system-wide ALSA configuration with Inferno-only setup for apps that have these environment variables changed.

This plugin is entirely user-space and contained in a library. It means that the Dante device is emulated only when the ALSA device is in use in an application and certainly can't outlive the process. When the stream is stopped or the whole audio app is closed, device disappears from the network. Next time it is opened, audio flows have to be established again and it takes time during which silence is played or recorded (in other words, several seconds of sound will be lost).

So the Inferno ALSA PCM is intended to be constantly running. The easiest way of ensuring this is using an audio server, for example JACK (not tested yet) or PipeWire (see script start_pipewire_sink), making sure that automatic suspending of audio device is disabled (or, to save energy, set to a timeout long enough that it won't be annoying). Some DAWs (e.g. Ardour, BespokeSynth, NOT Audacity) also keep the audio interface running all time.

• No matter whether the app does only capture, only playback or both, Inferno will always emulate both capture and playback device (unless number of channels is set to 0). If the app isn't playing audio, Dante devices receiving from Inferno will report broken stream (🚫 in Dante Controller) because, to save CPU, transmitting thread is not run at all.
• By default, only a single process can run Inferno because it listens on UDP sockets. To overcome this limitation, specify INFERNO_ALT_PORT which will be used as a start of the range of ports. Also specify INFERNO_DEVICE_ID and INFERNO_NAME (see main README for details) to avoid non-unique identifiers (Dante Controller may suffer from double vision in such cases).