ADF example pipeline_raw_http 441.k/24b issue

[micbel]

Hello everyone.

I've been trying different examples from esp-adf to get a feel of audio performance of esp32. I procured a LyraT v4.3 to conduct my experiments with. I was able to run pipeline_passthru example with I2S peripheral setup to 44.1k/24b which is a sampling freq/bit depth I'd like to use in future projects. I also tried running pipeline_raw_http example. It worked fine with standard settings, i.e. 16k/16b but .wav files captured with 44.1k/24b are just noise.

I played a bit with different settings and it seems like the issue is 24b encoding. When I try 44.1k/16b everything works just fine, but 16k/24b gives exact same results as 44.1k/24b. However when I take 44.1k/24b .wav file and modify the header to interpret the file content as 32b and 33.075k (which is 2/3 of 44.1k) you can clearly hear what was recorded, although pitch-shifted and with some noise still present.

Those experiments lead me to believe that I2S peripheral works fine (all signals look correct on the oscilloscope) and that there's an issue with (http)->(python server)->(wav encoding) part of the demo app but I'm having a hard time finding the exact cause. Can anyone shed some light? Have anyone tried this example with different than standard settings?

[famlee]

Hi,

Thank you for your good description. I believe that ESP32 has huge potential to play good quality audio but I feel the understanding of how to use it through examples and any description is very limited. I have been testing LYRAT board as well and feel that I have been stuck even earlier than you although it might be similar problem.

Here is a link to my issue: https://www.esp32.com/viewtopic.php?f=20&t=10811

As I have submitted the example of the playback so perhaps you can see whether it was similar issue to what you had and maybe this can shed some light on the bitrate settings that might be getting mixed in yours and our project.

Good luck with you problem and as soon as ours is solved I will post.