Sample Rate Accuracy

[frohro]

Hi @pschatzmann ,
I have been testing the sample rate accuracy using the WS clock of the I2S. I find it to be 48.333 kHz when it should be 48 kHz. When I try the Arduino-Pico code by itself, it gives a 48.046 kHz rate, which is almost an order of magnitude better. The reason I am concerned is that I am transferring the audio data to a computer and it is expecting 48 kHz. It will resample it if it has to, but the further off the frequencies are, the poorer my audio results will be. I thought you would be using the Arduino-Pico I2S, but it seems you are not. I also played around with changing the Raspberry Pi Pico clock. Your code changes the rate proportionally to the Pico clock, but the Arduino-Pico rate is independent of the Pico clock. Am I doing something wrong and/or what is the recommended way to fix this? I have considered changing the Pico crystal from 12 MHz to 12.288 MHz, but I'm hoping some software solution will be adequate. I could try using a Pico frequency of 124.2 MHz instead of 125 and that would bring me mich closer to 48 kHz. Would that have other effects I should be aware of?
Thanks,
Rob

[pschatzmann]

I don't quite understand: If you are using the RP2040 core from Earlee Phil Hower, then I am also just using the I2S API provided by the core!

[frohro]

I have tested both your implementation using the RP2040 core code and Earl's. Earl gives me a closer sample rate than the core. I would like at least as accurate a sample rate as Earl and better if possible. I guess that means that the core isn't as good as Earl's. Do I understand correctly? Do you see a problem with changing my Pico frequency to 124.2 MHz. It is one of the allowed values. I haven't tried defining RP2040_HOWER yet. Maybe I should. When I tried Earl's code it was without AudioTools at all.

[pschatzmann]

I would recommend Earle'core with my library: The Arduino core implementation provides only a limited functionality...

I don't really have a lot of experience with the RP2040 since I mainy do my testing on the available ESP32 variants..