ESP32 Forum

Hi,

When we use the DEFAULT_AAC_DECODER_CONFIG in audio pipelines, the audio is very clear. But when we use DEFAULT_ESP_M4A_DECODER_CONFIG in esp_audio with auto-decoder-detection and the track uses two channels, there is a background noise (like scratching/popping backgournd noise) in one of the channels only. Have you come across a similar issue? What is the deference between these two decoder implementations?

Also, is there a way to force DEFAULT_AAC_DECODER_CONFIG using esp_audio library, without auto decoder detection, on files without extension?

Thank you in advance,
Christos

A piece of information that might be helpful, DEFAULT_ESP_AAC_DECODER_CONFIG with auto decoder does not recognise our tracks.

The mp3 decoder has issues running on core 1 - it's fpu thing. Guessing M4A has the same thing?

Flac has no problem on core-1, mp3 decodes badly with noise.

It could be that? try running on core-0 to test.

Thank you shabtronic!

Switching the auto-decoder (esp_decoder_cfg_t) to core 0 actually did the trick!

Do you have more info/links on this matter? I cannot find any references online on Core 1 and codec issues.

Awesome!

No I have no links - I found it by myself when playing around with mp3/flac decoders on the different cores. I think it's linked to the task context switcher not saving the fpu state properly - that's just a guess tho - I didn't have time to investigate it further.