Some issues I have with the SD player

[AlphaGaming7780]

Hi,

I'm not opening a GitHub Issue because this is likely my own mistake, but I can't quite figure out why, and I also want to group multiple issues into one post.

1. I’m not entirely sure, but could the problem be that the copy function runs on the other core of the ESP32?

assert failed: T& audio_tools::Vector<T>::operator[](int) [with T = char] Vector.h:257 (index < len)


Backtrace: 0x40083aa1:0x3ffe9f80 0x40096499:0x3ffe9fa0 0x4009b9a1:0x3ffe9fc0 0x400d7283:0x3ffea0f0 0x400d73c6:0x3ffea110 0x400d7685:0x3ffea140 0x400d7742:0x3ffea180 0x400d7865:0x3ffea1b0 0x400da866:0x3ffea1e0 0x400daa25:0x3ffea230 0x40257751:0x3ffea260 0x400d4679:0x3ffea280
  #0  0x40083aa1 in panic_abort at /home/runner/work/esp32-arduino-lib-builder/esp32-arduino-lib-builder/esp-idf/components/esp_system/panic.c:408
  #1  0x40096499 in esp_system_abort at /home/runner/work/esp32-arduino-lib-builder/esp32-arduino-lib-builder/esp-idf/components/esp_system/esp_system.c:137
  #2  0x4009b9a1 in __assert_func at /home/runner/work/esp32-arduino-lib-builder/esp32-arduino-lib-builder/esp-idf/components/newlib/assert.c:85
  #3  0x400d7283 in audio_tools::Vector<char>::operator[](int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioBasic/Collections/Vector.h:257 (discriminator 1)
  #4  0x400d73c6 in audio_tools::MetaDataID3V2::processnotifyAudioChange() at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioMetaData/MetaDataID3.h:522
  #5  0x400d7685 in audio_tools::MetaDataID3V2::processTagNotFound(unsigned char const*, unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioMetaData/MetaDataID3.h:448
  #6  0x400d7742 in audio_tools::MetaDataID3V2::processPartialTagAtTail(unsigned char const*, unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioMetaData/MetaDataID3.h:495
      (inlined by) audio_tools::MetaDataID3V2::write(unsigned char const*, unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioMetaData/MetaDataID3.h:360
      (inlined by) audio_tools::MetaDataID3V2::write(unsigned char const*, unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioMetaData/MetaDataID3.h:353
      (inlined by) audio_tools::MetaDataID3::write(unsigned char const*, unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioMetaData/MetaDataID3.h:582
  #7  0x400d7865 in audio_tools::AudioPlayer::decodeMetaData(void*, void*, unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioPlayer.h:523
  #8  0x400da866 in audio_tools::StreamCopyT<unsigned char>::copyBytes(unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/StreamCopy.h:166 (discriminator 1)
  #9  0x400daa25 in audio_tools::AudioPlayer::copy(unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioPlayer.h:373
      (inlined by) audio_tools::AudioPlayer::copy(unsigned int) at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioPlayer.h:360
  #10 0x40257751 in audio_tools::AudioPlayer::copy() at .pio/libdeps/PortalRadio/audio-tools/src/AudioTools/CoreAudio/AudioPlayer.h:356
  #11 0x400d4679 in AudioPlayerLoopTask(void*) at src/Audio/AudioManager.cpp:254 (discriminator 1)

2. When an audio file contains metadata, the first second of playback produces strange noise, almost as if the stream is restarting. If you need an audio file to reproduce this issue, let me know.

3. What’s the proper way to switch from one audio source to another, especially from SD to another source? Sometimes, when switching from SD, the ESP32 freezes—the code stops running, and I have to restart it. Other sources work fine, but SD doesn’t. I'm simply killing the task running AudioPlayer.copy in a loop on the second core, then calling AudioPlayer.End(), but sometimes the code gets stuck there. I'm using MP3DecoderHelix as the decoder.

4. It would be great if SDSouce.End() didn't necessarily terminate the SD instance, in case it's still needed.

Thanks in advance for any insights!

[pschatzmann]

1+2) If you don't use the metadata, I recommend to filter it out using a MetaDataFilerDecoder. Since the amount of RAM is limited it is assumed that the file only has reasonably sized metadata.

Please provide the example file that is cusing the crash: My first guess would be that you just run out of RAM.
That would match with with our other issues.

3. My guess would be:

• call end on the player
• set the new source with setAudioSource
• call begin()
• make sure that no copy() calls are executed while you are doing the above.

4. Well: just don't call end() if you don't want to close the SD instance: that is all the end() is doing!

[pschatzmann]

I don't quite understand: the MP3 decoder does not need any time. It is important that it gets closed and reopened, so that no stale data is availble from the previous source, but this is taken care of by calling end and begin on the player.

This should take millisenconds but never 5 seconds!

After switching, you can dynamically phase in the volume e.g. like this:

for (int j=0;j<100;j++){
volume.setVolume(100.0/j);
copier.copy();
}

[AlphaGaming7780]

Yeah, but why it's not taking a few milliseconds, if you watch the video, you will see it's doing a bunch of stream.copy whitout playing any song, and in the middle of these stream copy, the MP3 decoder is reconfiguring it self wich make weird noise.

[pschatzmann]

This would only happen when your new mp3 is damaged or the decoder has not been closed before sending some new data.
Did you try to install the latest version of libhelix ?

[AlphaGaming7780]

I'm using version 0.8.7 of libhelix.

The decoder is closed when we do AudioPlayer.end() ?, if yes, I'm closing the decoder before setting a new source.

How can know if my MP3 is damaged ?, it's playing nice on my PC and after the weird noise it's playing nice too, it's just take 5 sec before it start playing.

[pschatzmann]

Just test it with a separate simple sketch.

I already gave you the solution approach how you could suppress this noise....