tiny "pop" sound after fading sound

[KatePimbleton]

Hello,

I'm using merge, it takes 2 mono inputs, merge in a stereo output, copy to I2S.
When 1 channel is playing and I fade the other ( fade in or fade out ) everything
works fine.
My problem is when I fade a channel while the other is already muted, in that
case after the fade out or before fade in I can here a tiny tiny "pop" sound.
What strategy should I employ to avoid this ?

Cheers

[pschatzmann]

Not sure what the issue exactly is, but I also noticed a pop when we start to send any data to I2S.
If this is the cause of your issue, then a work around would be just to send silence...

[KatePimbleton]

I think this is not the cause, I'm already sending data constantly to I2s,
I did further testing, playing sound continuously ( fileloop1 fileloop2 -> merge -> volume -> i2s
I'm controlling I2s volume by OSC from my computer,
whenever It passes above or under 0.005 I get a pop.

I'll do other tests

[KatePimbleton]

I made a simplified version of my sketch, the test sound I use is an composed of sine and silence alternating.
for each silence moment there's a pop at beginning and end.

( also if I don't add these lines
uint8_t testBuffer[64];
decoder.readBytes(testBuffer, 64);
in the setup no sound is coming out at all, no idea why,maybe related? )

#include "FS.h"
#include <SD_MMC.h>
#include "AudioTools.h"
#include "AudioTools/AudioLibs/AudioBoardStream.h"
#include "AudioTools/Disk/FileLoop.h"

using namespace audio_tools;

AudioInfo info(44100, 2, 16);
AudioBoardStream i2s(AudioKitEs8388V1);
InputMerge<int16_t> merge;

FileLoop loopingFile;
EncodedAudioStream decoder(&loopingFile, new WAVDecoder());

StreamCopy copier(i2s, merge, 512);

File file1;

void setup() {
Serial.begin(115200);

auto config = i2s.defaultConfig(TX_MODE);
config.copyFrom(info);
config.sd_active = true;
i2s.begin(config);

SD_MMC.begin("/sdcard", true, true, SDMMC_FREQ_52M);

merge.add(decoder, 1);
merge.begin(infoMerge);
decoder.begin();

file1 = SD_MMC.open("/tones/toneAndSilence.wav");
loopingFile.setFile(file1);
loopingFile.begin();

uint8_t testBuffer[64];
decoder.readBytes(testBuffer, 64);
}

void loop() {
copier.copy();
}

[pschatzmann]

Can you share you test file as well ?

[KatePimbleton]

toneAndSilence.wav.gz

[KatePimbleton]

we hear 2 clicks when going from sound to silence, first one is because of sudden volume and can be ignored

[KatePimbleton]

bypassing merge does the same atcually:

#include "FS.h"
#include <SD_MMC.h>
#include "AudioTools.h"
#include "AudioTools/AudioLibs/AudioBoardStream.h"
#include "AudioTools/Disk/FileLoop.h"

using namespace audio_tools;

AudioBoardStream i2s(AudioKitEs8388V1);

FileLoop loopingFile;
EncodedAudioStream decoder(&loopingFile, new WAVDecoder());

StreamCopy copier(i2s, decoder, 512);

File file1;

void setup() {
Serial.begin(115200);
auto config = i2s.defaultConfig(TX_MODE);
config.copyFrom(decoder.audioInfo());
config.sd_active = true;
i2s.begin(config);
SD_MMC.begin("/sdcard", true, true, SDMMC_FREQ_52M);

file1 = SD_MMC.open("/tones/toneAndSilence.wav");
loopingFile.setFile(file1);
loopingFile.begin();

decoder.begin();

}

void loop() {
copier.copy();
}

[KatePimbleton]

using an arduino audio kit v2.2 3378

[pschatzmann]

Have a look and play your file in Audacity: the issue is in your data since I can reproduce it there as well!

You should fade in and fade out: at least your samples should start/end at 0...

[KatePimbleton]

toneAndSilence2.wav.gz
you're right my test sound is ambiguous,
here is a faded version of the test sound so you only hear the tiny "pop" sound whenever it goes to or from silence.

[KatePimbleton]

I will test on another esp32 with external dac to check if that comes from the esp32-audiokit hardware

[Queso-su]

My dev :
esp32c3
max98357a
code from sin example

No change everything , pup voice same.
but when I use deepseek rec code , its work ,
I dont know why
mycode :

MAX98375A

#include <Arduino.h>
#include <driver/i2s.h>

class MAX98375A_Output {
private:
    int bclkPin;
    int lrckPin;
    int dinPin;
    float currentFrequency;
    float currentVolume;
    bool isPlaying;
    TaskHandle_t audioTaskHandle;

    static void audioTask(void* parameter) {
        MAX98375A_Output* instance = (MAX98375A_Output*)parameter;
        
        // I2S配置
        i2s_config_t i2s_config = {
            .mode = (i2s_mode_t)(I2S_MODE_MASTER | I2S_MODE_TX),
            .sample_rate = 44100,
            .bits_per_sample = I2S_BITS_PER_SAMPLE_16BIT,
            .channel_format = I2S_CHANNEL_FMT_RIGHT_LEFT,
            .communication_format = I2S_COMM_FORMAT_STAND_I2S,
            .intr_alloc_flags = ESP_INTR_FLAG_LEVEL1,
            .dma_buf_count = 8,
            .dma_buf_len = 64
        };

        i2s_pin_config_t pin_config = {
            .bck_io_num = instance->bclkPin,
            .ws_io_num = instance->lrckPin,
            .data_out_num = instance->dinPin,
            .data_in_num = I2S_PIN_NO_CHANGE
        };

        // 安装I2S驱动程序
        i2s_driver_install(I2S_NUM_0, &i2s_config, 0, NULL);
        i2s_set_pin(I2S_NUM_0, &pin_config);

        const int bufferSize = 1024;
        int16_t* buffer = (int16_t*)malloc(bufferSize * sizeof(int16_t));

        float phase = 0;
        
        while (instance->isPlaying) {
            // 实时获取频率和音量
            float freq = instance->currentFrequency;
            float vol = instance->currentVolume;
            
            if (freq > 0) {
                // 生成音频数据
                float phaseIncrement = 2 * PI * freq / 44100;
                for (int i = 0; i < bufferSize; i += 2) {
                    buffer[i] = sin(phase) * 32767 * vol;
                    buffer[i + 1] = buffer[i];
                    phase += phaseIncrement;
                    if (phase >= 2 * PI) {
                        phase -= 2 * PI;
                    }
                }
                
                // 输出音频
                size_t bytesWritten;
                i2s_write(I2S_NUM_0, buffer, bufferSize * sizeof(int16_t), &bytesWritten, portMAX_DELAY);
            } else {
                // 频率为0，生成静音
                memset(buffer, 0, bufferSize * sizeof(int16_t));
                size_t bytesWritten;
                i2s_write(I2S_NUM_0, buffer, bufferSize * sizeof(int16_t), &bytesWritten, portMAX_DELAY);
                vTaskDelay(10 / portTICK_PERIOD_MS);
            }
        }

        // 清理资源
        free(buffer);
        i2s_driver_uninstall(I2S_NUM_0);
        vTaskDelete(NULL);
    }

public:
    MAX98375A_Output(int bclk = 4, int lrck = 5, int din = 9)
        : bclkPin(bclk), lrckPin(lrck), dinPin(din),
          currentFrequency(0.0), currentVolume(1.0), isPlaying(false) {}

    void begin() {
        if (!isPlaying) {
            isPlaying = true;
            xTaskCreatePinnedToCore(audioTask, "AudioTask", 4096, this, 1, &audioTaskHandle, 1);
        }
    }

    void stop() {
        isPlaying = false;
        // 等待任务结束
        if (audioTaskHandle != NULL) {
            vTaskDelay(100 / portTICK_PERIOD_MS); // 给任务一些时间结束
        }
        currentFrequency = 0;
    }

    void setFrequency(float freq) {
        currentFrequency = freq;
        // 如果频率有效且当前没有播放，则开始播放
        if (freq > 0 && !isPlaying) {
            begin();
        }
    }

    void setVolume(float volume) {
        if (volume < 0.0) volume = 0.0;
        if (volume > 1.0) volume = 1.0;
        currentVolume = volume;
    }

    float getFrequency() {
        return currentFrequency;
    }

    float getVolume() {
        return currentVolume;
    }

    bool isActive() {
        return isPlaying;
    }
};

Main

#include <Arduino.h>
#include <math.h>
#include <driver/i2s.h> 

#include "MAX98375A_Output.h"

#include "AudioTools.h"

AudioInfo info(44100, 2, 16);
SineFromTable<int16_t> sineWave(5000);                // subclass of SoundGenerator with max amplitude of 32000
            // subclass of SoundGenerator with max amplitude of 32000
GeneratedSoundStream<int16_t> sound(sineWave);             // Stream generated from sine wave
I2SStream out; 
StreamCopy copier(out, sound);       

// 矩阵键盘引脚定义
const int rowPins[3] = {19, 18, 1};    // 行引脚
const int colPins[4] = {7, 6, 10, 3};  // 列引脚

// 按键映射
const char keyMap[3][4] = {
  {'5', '6', '7', 'A'},
  {'2', '3', '4', '1'},
  {'C', 'D', 'E', 'B'}
};

// 音符频率定义 (单位: Hz)
#define NOTE_C4  262  // 中央C
#define NOTE_CS4 277  // C#
#define NOTE_D4  294  // D
#define NOTE_DS4 311  // D#
#define NOTE_E4  330  // E
#define NOTE_F4  349  // F
#define NOTE_FS4 370  // F#
#define NOTE_G4  392  // G
#define NOTE_GS4 415  // G#
#define NOTE_A4  440  // A
#define NOTE_AS4 466  // A#
#define NOTE_B4  494  // B
#define NOTE_C5  523  // 高音C

// I2S配置 - 用于MAX98357A
#define I2S_BCK_PIN 4
#define I2S_WS_PIN 5
#define I2S_DOUT_PIN 9



// 音频参数
#define SAMPLE_RATE 22050
#define AMPLITUDE 8000  // 降低振幅
#define BUFFER_SIZE 256  // 较小的缓冲区


// 按键状态跟踪
bool keyStates[3][4] = {{false}};
bool lastKeyStates[3][4] = {{false}};
unsigned long debounceTimes[3][4] = {{0}};
const unsigned long debounceDelay = 20;

// 当前播放的音符
int currentNote = 0;
bool isPlaying = false;


MAX98375A_Output audioOutput(4, 5, 9);

  


// 根据字符获取音符频率
int getNoteFrequency(char key) {
  switch(key) {
    case '1': return NOTE_C4;   // C4
    case '2': return NOTE_D4;   // D4
    case '3': return NOTE_E4;   // E4
    case '4': return NOTE_F4;   // F4
    case '5': return NOTE_G4;   // G4
    case '6': return NOTE_A4;   // A4
    case '7': return NOTE_B4;   // B4
    case 'A': return NOTE_CS4;  // C#4
    case 'B': return NOTE_DS4;  // D#4
    case 'C': return NOTE_FS4;  // F#4
    case 'D': return NOTE_GS4;  // G#4
    case 'E': return NOTE_AS4;  // A#4
    default: return 0;
  }
}



void setup() {
  copier.begin();
  Serial.begin(115200);
  Serial.println("电子琴初始化...");
   while(!Serial);
  AudioToolsLogger.begin(Serial, AudioToolsLogLevel::Info);

  // start I2S
  Serial.println("starting I2S...");
  auto config = out.defaultConfig(TX_MODE);
  config.pin_bck = 4;
  config.pin_ws = 5;
  config.pin_data = 9;
  config.copyFrom(info); 
  out.begin(config);

  // Setup sine wave
  sineWave.begin(info, N_B4);
  Serial.println("started...");

  // 初始化矩阵键盘引脚
  for (int i = 0; i < 3; i++) {
    pinMode(rowPins[i], OUTPUT);
    digitalWrite(rowPins[i], HIGH);
  }
  
  for (int j = 0; j < 4; j++) {
    pinMode(colPins[j], INPUT_PULLUP);
  }
  

  
}
void loop() {
   copier.copy();
  static int lastNote = 0;  // 用于记录上一次按下的音符，避免重复设置相同的频率
  int currentNote = 0;
  bool anyKeyPressed = false;
  
  // 扫描所有按键
  for (int row = 0; row < 3; row++) {
    digitalWrite(rowPins[row], LOW);
    
    for (int col = 0; col < 4; col++) {
      bool currentState = (digitalRead(colPins[col]) == LOW);
      
      if (currentState != lastKeyStates[row][col]) {
        debounceTimes[row][col] = millis();
      }
      
      if ((millis() - debounceTimes[row][col]) > debounceDelay) {
        if (currentState != keyStates[row][col]) {
          keyStates[row][col] = currentState;
          
          if (currentState) {
            char keyChar = keyMap[row][col];
            int note = getNoteFrequency(keyChar);
            if (note > 0) {
              Serial.print("按键按下: ");
              Serial.print(keyChar);
              Serial.print(" - 频率: ");
              Serial.println(note);
              currentNote = note;
            }
          }
        }
      }
      
      lastKeyStates[row][col] = currentState;
      
      // 检查是否有按键按下
      if (keyStates[row][col]) {
        anyKeyPressed = true;
      }
    }
    
    digitalWrite(rowPins[row], HIGH);
    delay(1);
  }
  
  // 处理音频播放
  if (anyKeyPressed) {
    if (currentNote != 0) {
      // 只有当音符改变时才更新频率，避免重复设置
      if (currentNote != lastNote) {
        audioOutput.setFrequency(currentNote);
        audioOutput.setVolume(0.1);
        lastNote = currentNote;
      }
    }
  } else {
    // 没有按键按下，设置频率为0（静音）
    if (lastNote != 0) {
      delay(100); // 确保音符完全停止
      audioOutput.setFrequency(0);
      lastNote = 0;
    }
  }
  
  delay(5); // 主循环延迟
}

[pschatzmann]

I had a closer look into this and found that the dynamic switching from mono to stereo is not working: Since there is no connection between the encoder and I2S a manual notification registration is needed!

I edited your second wav file my adding some more silence in the beginning and the issue has dissapeared. However I don't quite understand why it was there in the first place: Maybe the gap between the last sample and the first was too big.

Here is the improved sketch that plays with mono now

#include "FS.h"
#include <SD_MMC.h>
#include "AudioTools.h"
#include "AudioTools/AudioLibs/AudioBoardStream.h"
#include "AudioTools/Disk/FileLoop.h"

AudioBoardStream i2s(AudioKitEs8388V1);
//CsvOutput out(Serial,1);
File file1;
FileLoop loopingFile;
WAVDecoder wav;
EncodedAudioStream decoder(&loopingFile, &wav);
StreamCopy copier(i2s, decoder);

void setup() {
  Serial.begin(115200);
  AudioToolsLogger.begin(Serial, AudioToolsLogLevel::Warning);

  auto config = i2s.defaultConfig(TX_MODE);
  i2s.begin(config);

  SD_MMC.begin("/sdcard");
  file1 = SD_MMC.open("/wav-test/toneAndSilence3.wav");
  loopingFile.setFile(file1);
  loopingFile.begin();

  decoder.addNotifyAudioChange(i2s);
  decoder.begin();
}

void loop() {
  copier.copy();
}

With the improved file: toneAndSilence3.wav

[KatePimbleton]

I tested the esp32 audiokit I use with aoo for esp32 ( https://git.iem.at/ressi/esp32_aoo_test ), the same thing happens, when level goes to zero it produces a tiny "puup" , and another one when it goes from silence to sound.
This sound is unnoticeable if I set i2s volume under 0.6, which is ok for my project.

offtopic:
I'm used to arduino audio tools now, it would be great to have an esp32aoo reception implementation!