A2DP stream sounds weird

[UnforeseenOcean]

Code (it is a spaghetti code, but it's a basis for future changes):

#include <OneButton.h>
#include "AudioTools.h"
#include "AudioTools/AudioLibs/AudioBoardStream.h" // install https://github.com/pschatzmann/arduino-audio-driver
#include "AudioTools/AudioCodecs/CodecMP3Helix.h"
#include "BluetoothA2DPSink.h"
#include "lamesound2.h"

#define KEY_VOLUP 18
#define KEY_NEXT 23
#define KEY_PLAY 19
#define KEY_PREV 13
#define KEY_VOLDN 36
#define BATT_PIN 15

uint8_t curVolume = 63;
bool isMuted = false;
uint16_t battTick = 0;
uint16_t battCheckTick = 30000;

AudioBoardStream kit(AudioKitEs8388V1);
BluetoothA2DPSink a2dp_sink(kit);
MemoryStream mp3;
MP3DecoderHelix helix;
EncodedAudioStream out(&kit, &helix);
StreamCopy copier(out, mp3);

OneButton volup_btn(KEY_VOLUP, true);
OneButton next_btn(KEY_NEXT, true);
OneButton play_btn(KEY_PLAY, true);
OneButton prev_btn(KEY_PREV, true);
OneButton voldn_btn(KEY_VOLDN, true);

void playSound(int index) {
  const uint8_t* data = nullptr;
  uint32_t data_len = 0;
  ESP_LOGI("SoundPlay", "Playing sound %d", index);
  switch (index) {
    case 0: {
      data = battlow;
      data_len = battlow_len;
      break;
    }
    case 1: {
      data = connected;
      data_len = connected_len;
      break;
    }
    case 2: {
      data = disconnected;
      data_len = disconnected_len;
      break;
    }
    case 3: {
      data = initSnd;
      data_len = initSnd_len;
      break;
    }
    case 4: {
      data = linkloss;
      data_len = linkloss_len;
      break;
    }
    case 5: {
      data = volmax;
      data_len = volmax_len;
      break;
    }
    case 6: {
      data = volmin;
      data_len = volmin_len;
      break; 
    }
    case 7: {
      data = poweron;
      data_len = poweron_len;
      break;
    }
    default: {
      data = poweroff;
      data_len = poweroff_len;
      break;
    }
  }
  ESP_LOGW("SoundPlay", "Init data, len = %u", data_len);

  if (data == nullptr || data_len == 0) {
    ESP_LOGE("SoundPlay", "Something is wrong, wrong or empty data supplied.!!");
    return; 
  }

  mp3.setValue(data, data_len);
  mp3.begin();
  helix.setOutput(kit);
  ESP_LOGW("SoundPlay", "Start play...");
  helix.begin();
  ESP_LOGW("SoundPlay", "Enter loop");
  while (mp3.available()) {
    copier.copy();
  };
  ESP_LOGW("SoundPlay", "Exit loop, destroy Helix decoder...");
  helix.end();
  ESP_LOGW("SoundPlay", "Freeing reserved memory for MemoryStream...");
  mp3.end();
  
  ESP_LOGW("SoundPlay", "Play sound %d ok...", index);
}

void goNext() {
  ESP_LOGI("OneButton", "NEXT action initiated");
  a2dp_sink.next();
}

void goPlay() {
  ESP_LOGI("OneButton", "PLAY action initiated");
  if (a2dp_sink.is_output_active() == true) {
    a2dp_sink.pause();
  } else {
    a2dp_sink.play();
  }
}

void goPrev() {
  ESP_LOGI("OneButton", "PREV action initiated");
  a2dp_sink.previous();
}

void mute() {
  ESP_LOGI("OneButton", "MUTE action initiated");
  if (isMuted) {
    ESP_LOGI("VolumeMgr", "Volume set to %d", curVolume);
    a2dp_sink.set_volume(curVolume);
  }
  else {
    ESP_LOGI("VolumeMgr", "Volume set to 0");
    curVolume = a2dp_sink.get_volume(); // Update current volume with actual volume
    a2dp_sink.set_volume(0);
  }
  isMuted = !isMuted;
}
void connectionStatusChanged(esp_a2d_connection_state_t state, void *ptr) {
  switch (state) {
    case esp_a2d_connection_state_t::ESP_A2D_CONNECTION_STATE_DISCONNECTED:
      ESP_LOGI("Bluedroid", "A2DP Source Disconnected");
      playSound(2);
      break;
    case esp_a2d_connection_state_t::ESP_A2D_CONNECTION_STATE_CONNECTED:
    ESP_LOGI("Bluedroid", "A2DP Source Connected");
      playSound(1);
      break;
    default:
      // Nothing
      break;
  }
}

void volumeUpFine() {
  curVolume = a2dp_sink.get_volume();
  curVolume = curVolume + 1;
    if (curVolume > 127) {
    curVolume = 127;
    ESP_LOGI("VolumeMgr", "Volume is at 127");
  }
  ESP_LOGI("VolumeMgr", "Volume set to %u", curVolume);
  a2dp_sink.set_volume(curVolume);
}

void volumeUpCoarse() {
  curVolume = a2dp_sink.get_volume();
  curVolume = curVolume + 4;
  if (curVolume > 127) {
    curVolume = 127;
    ESP_LOGI("VolumeMgr", "Volume is at 127");
  }
  ESP_LOGI("VolumeMgr", "Volume set to %u", curVolume);
  a2dp_sink.set_volume(curVolume);
}

void volumeDownFine() {
  curVolume = a2dp_sink.get_volume();
  curVolume = curVolume - 1;
  if (curVolume <= 0) { 
    curVolume = 0;
    ESP_LOGI("VolumeMgr", "Volume is at 0");
  }
  ESP_LOGI("VolumeMgr", "Volume set to %u", curVolume);
  a2dp_sink.set_volume(curVolume);
}

void volumeDownCoarse() {
  curVolume = a2dp_sink.get_volume();
  curVolume = curVolume - 4;
  if (curVolume <= 0) { 
    curVolume = 0;
    ESP_LOGI("VolumeMgr", "Volume is at 0");
  }
  ESP_LOGI("VolumeMgr", "Volume set to %u", curVolume);
  a2dp_sink.set_volume(curVolume);
}

void setupKeys() {
  ESP_LOGI("OneButton", "Begin setting up buttons...");
  // Setup buttons
  ESP_LOGI("OneButton", "Attach button int...");
  volup_btn.attachClick(volumeUpFine);
  volup_btn.attachDuringLongPress(volumeUpCoarse);
  next_btn.attachClick(goNext);
  play_btn.attachClick(goPlay);
  prev_btn.attachClick(goPrev);
  voldn_btn.attachClick(volumeDownFine);
  voldn_btn.attachDuringLongPress(volumeDownCoarse);

  // Set delays
  ESP_LOGI("OneButton", "Set key debounce...");
  volup_btn.setDebounceMs(100);
  next_btn.setDebounceMs(100);
  play_btn.setDebounceMs(100);
  prev_btn.setDebounceMs(100);
  voldn_btn.setDebounceMs(100);

  ESP_LOGI("OneButton", "Set key duration...");
  volup_btn.setClickMs(500);
  next_btn.setClickMs(500);
  play_btn.setClickMs(500);
  prev_btn.setClickMs(500);
  voldn_btn.setClickMs(500);

  ESP_LOGI("OneButton", "Set key hold duration");
  volup_btn.setPressMs(1000);
  next_btn.setPressMs(1000);
  play_btn.setPressMs(1000);
  prev_btn.setPressMs(1000);
  voldn_btn.setPressMs(1000);
  ESP_LOGI("OneButton", "Key setup OK?");
}

void setup() {
  Serial.begin(115200);
  auto cfg = kit.defaultConfig();
  cfg.sample_rate = 44100;
  cfg.channels = 2;
  kit.begin(cfg);
  out.begin();
  infoDump();
  setupKeys();
  logWhodunit();
  playSound(7);
  a2dp_sink.set_on_connection_state_changed(connectionStatusChanged);
  a2dp_sink.start("ESP-BT-DRG");
  logWhodunit();
  ESP_LOGI("System", "Ready");
}

void loop() {
  volup_btn.tick();
  next_btn.tick();
  play_btn.tick();
  prev_btn.tick();
  voldn_btn.tick();
}

The prompt sound works, however the moment the sound plays through the Bluetooth connection, the sound is low in pitch and the CPU gets unstable. I'm pretty sure I did something wrong, and I need some help figuring out why.

Potential answers:

• Wrong sampling rate (A2DP and MP3 Helix has different sampling rate somehow)
• Wrong channel count (It currently uses "2" but it was "1" in the example. The MP3 files are mono.)
• Some part of the code putting way too much strain on the processor, causing buffer underrun

[pschatzmann]

Just store the actual AudioInfo and restore it back to the original after you are done playing your file.

[UnforeseenOcean]

An overview on how one could do that would've been good. We all know you're not doing this full-time, so you probably are very preoccupied, thus asking for a code would be asking too much (we all know what happened when I did that). Hence, I only ask for something like this:

1. Process A begins
2. While process A is ongoing, do process B
3. Wait until process A is done
4. With the info from process B, do process C
5. Done

[UnforeseenOcean]

<< PLEASE READ FULLY BEFORE ANSWERING. Thank you for your cooperation. >>

I hate to say this because I might come off as being rude, but please do not mark your answer as an "answer" if all you'll tell me is one line of clue as if I'm in a round of 20 Questions.

This is seriously frustrating. I'm trying to learn, and yet you just tell me that "just do this" without ever telling me how. I'm not a native English speaker and it takes me a while to understand the sentences. Doing this does not help the matter one bit.

I'm not insulting you or your integrity. Most of what you said just flies right over my head despite my best efforts, yet I am trying my hardest to try to fix my code before asking questions. What you don't see from the discussion threads or issue report is hundreds and thousands of debugging attempts I made.

I am very certain that what you made is genuinely amazing and I could never dream of writing such clean code as you did. I actually recommended your libraries for others to use whenever the occasion occurs. They're delightful when they work, and makes it very easy to do complex things. With that said, I'm simply telling you that this act of answering the question with a one-liner then immediately marking it as solved isn't helping others as much as you think it does.

Regardless, I thank you for your hard work, and I hope you have a nice day.

[UnforeseenOcean]

I have tried few fixes that were suggested, and here's the result:

The audio was stuttering and low in pitch before, with a long delay between play and pause (playing the audio and stopping the playback when I tell it to stop) before the change.

• Try different sampling rate, specifically that of A2DP output
  • Irrelevant, both are at 44100Hz and changing the code made no changes
• Change event stack size and queue size
  • Causes crash upon connection
• Change buffer size and count
  • Makes no difference, sound still garbled
• Use BluetoothA2DPSinkQueued
  • Fixes the sound
  • Causes other things to fail to be allocated because of a huge memory footprint
  • Things like LEDs, animations for ARGB beads, prompt sounds and external control task will fail to initialize after connection is made

After the change, the sound is fixed but it's taking up too much memory.

[UnforeseenOcean]

Setting the a2dp_sink.seti2s_ringbuffer_size to 32768 (the supposed default size) causes ringbuffer is full, drip this packet! error and ringbuffer overflowed, ready to decrease data! mode changed: RINGBUFFER_MODE_DROPPING (which could be the reason for the weird low-pitched audio)

[UnforeseenOcean]

In fact, just defining that causes constant faults that result in the audio sounding as if it's being played through a bubble wrap. I need other ways of optimizing it so I can fit everything together.

[UnforeseenOcean]

Alright, I don't even know what's happening anymore. Somehow, it worked fine, then it just refuses to work correctly. I even told it to erase its flash memory fully, to no avail.

[UnforeseenOcean]

How come setting the ring buffer size to three times its original intended size still result in "ring buffer full" error? I don't get it...

Experiments:

• ringbuffer_size = 1024: No audio, errors
• ringbuffer_size = 16384: Distorted audio, errors
• ringbuffer_size = 32768: Distorted audio, errors
• ringbuffer_size = 65536: Distorted audio, errors
• ringbuffer_size = 98304: Distorted audio, you know the drill

[UnforeseenOcean]

I think I partially know what's going on -- PSRAM settings. If it's turned on, no matter the configuration, it will always fail to work correctly.