First off, if you “just” want 16 channels of 16 bit audio, there are easier routes, e.g.
this thread.
16 channels of 16-bit audio at 44.1kHz is 11.29Mbps. Teensy 4.x can do that, on the face of it. But … looking at the data sheet you can see it’s not
really a SPI protocol, even if TI call it that; and it has an extra overhead of 16 bits for every sample. Now we’re at 22.58Mbps, which is still inside the Teensy’s SPI bus official limit of 30MHz. And you still have to persuade the LPSPI peripheral to generate the SYNC signal your DAC wants, which will require trying to understand the relevant part of the 3,500-page datasheet. Oh, and since you want two DACs, each one needs its own SYNC; that may or may not require you to use two SPI peripherals, but at least there are 3 on Teensy 4.x.
If that is possible, then the FIFO is very much relevant, although at only 16 words deep it’s still going to need refilling every 22us. That should be possible with DMA, there’s a bunch of code in the SPI driver, and also for asynchronous SPI displays, which you could review and adapt.
You will probably need to figure out how to adjust the SPI clock to get a sample rate of 44.1kHz - normally people don’t much care about the exact speed, and I think it’s an integer division of 240MHz by default, which you can’t really use.
Alternatively it may be possible to use the SAI audio peripheral to do what you want, if there’s some way of getting the word clock to provide the SYNC signal. In that case there’s lots of examples in the Audio library of driving hardware, and it has a 44.1kHz clock already.
