[3] Integrate ALSA Audio I/O buffering on Raspberry for an existing custom C++ audio project

I have my own fully custom C++ audio DSP project that currently runs on a dedicated DSP platform (Sharc ADSP). I want to migrate/convert this project to a VSC project on Raspberry Pi5. The C++ source code is already in a „Rpi-portable” version, i.e. stripped of Sharc peripherals-specific code Hardware concept: • Use of two USB to I2S/SPDIF UAC2 converters like this: [login to view URL] ((as input Converter1 and output Converter2) or [login to view URL] (as output Converter2) • I2S pins of Converter 1 will be connected to RPi GPIO header and feed input audio signal via the standard RPi I2S port • Converter 2 will be connected to a Rpi USB port and will output audio signal via ALSA USB audio device Scope of work - 3 main tasks: • The current project has two „interface” buffers – handling a single stereo input buffer and a stereo output buffer. Task 1): The new code should handle input and output data flow around the current project 1,1) Feed the I2S standard port incoming data to the X[] buffer of the current project 1,2) Signal an incoming audio sample rate change and pass the current SR value to one of my classes 1,3) Transfer out the data from the current Y[] post-DSP buffer to the ALSA UAC2 device • The project currently performs intensive DSP processing as a predefined sequence of tasks that should be set up in Linux Task 2): Project should be performance-optimized for Linux Rpi - probably making the DSP processing steps independent processing threads could bring better CPU utilization in Linux ? • The project uses IIR filters in DSP processing, currently a Sharc internal DSP library is used, a library optimized for ARM should be linked Task 3): Replace with an optimized ARM DSP library for 64-bit calculations, e.g. kfrlib Attachments: 1) A block diagram that shows the integration points between the current project and the input and output audio streams on RPi 2) Selected C++ "border" classes from the current Sharc project that need to be integrated with the input/output audio handling to be implemented. Key source files: - [login to view URL] - just the top first function is important - here all DSP processing is run (to be repolaced by threads) - [login to view URL] - the PACK_MAN::GETX_SPDIF24() function to be replaced with a call to your implementation that will fill the ALSA UAC2 output buffers I also have source code of a working RPI ALSA audio project that takes data from netwrork, applies DSP processing and then outputs to I2S buffers using ALSA. In direct talks I can share this project which can be a basis to create the required one and could speed up the implementation work.