I’m looking for a relatively powerful DSP & ML audio processing system for at least 8 I2S microphones. The 970 processor seems to fit the bill, but does the HiKey970 board break out all 8 I2S channels? I couldn’t find clear documentation on this.
I read the DSP documentation & driver is now available in Android, is it also available in Debian / plain Linux?
There is only a single i2s input and output available on the board (i.e. 2-channels/stereo in, and 2-channels out). The hardware is compatible with TDM, which would be able to transfer up to 10 channels in and 10 channels out, HOWEVER, to my knowledge, there is no kernel support for this.
Thanks for the pointer! I did a quick search and it seems the 4.12 ASoC subsystem supports TDM over PCM and the wording seems to indicate this is a high-level affair (no device specific support needed for decoding what the hardware generates): ASoC Digital Audio Interface (DAI) — The Linux Kernel documentation
Conceptually very simple, but in practice, not quite so. You do have to configure the hardware for it (DMAs and such). In any case, I suspect that there may be enough data in the public domain to accomplish this. Not that it will be easy. Look at the kernels for commercial release products (i.e. cellphones) with the kirin970. They will probably have the missing hardware setup pieces.
