2014-02-25, 23:41
Audio engine has a couple of stages where audio buffers can be processed. If nothing needs to be done the buffer is passed through the stage. Check out https://github.com/xbmc/xbmc/blob/master....cpp#L1604
We can prepare additional stages for manipulating audio buffers. The engine would check if any addon dsp has registered a function to a stage and call that. This interface has to be implemented once in C/C++.
This sounds trivial but we need more functions for configuration. Consider a case where the audio driver provides a sink with 48khz sample rate and the content is 44.1khz. Now we have to find out whether DSP or engine has to to resampling. So we need to define capabilities and addon can implement which can be queried for a given combination of sink, content, etc.
We can prepare additional stages for manipulating audio buffers. The engine would check if any addon dsp has registered a function to a stage and call that. This interface has to be implemented once in C/C++.
This sounds trivial but we need more functions for configuration. Consider a case where the audio driver provides a sink with 48khz sample rate and the content is 44.1khz. Now we have to find out whether DSP or engine has to to resampling. So we need to define capabilities and addon can implement which can be queried for a given combination of sink, content, etc.
