I2S external dac works with mono but not stereo signals

[awwende]

I've been playing with the MAX98357A (https://www.adafruit.com/product/3006). I've been using SparkFun's ESP32 Thing (https://www.sparkfun.com/products/13907) with the motion shield (https://www.sparkfun.com/products/14430) to play .wav files from an SD card. I used audacity to convert the stereo signals to mono and exported the file as a WAV signed 16-bit PCM. The mono track plays fine, but when I export the stereo track to the same format it will play, but there's a lot of noise on top of the actual track that's playing. I've dropped the sample rate down from 44100 to 22050 and the noise goes away, but of course the file plays at half speed.

I've been using the Arduino 1.8.5 IDE with the latest release from the arduino-esp32 github page with the following code:

Code: Select all

#include <SD.h>
#include "driver/i2s.h"
#include "freertos/queue.h"

#define CCCC(c1, c2, c3, c4)    ((c4 << 24) | (c3 << 16) | (c2 << 8) | c1)

/* these are data structures to process wav file */
typedef enum headerState_e {
    HEADER_RIFF, HEADER_FMT, HEADER_DATA, DATA
} headerState_t;

typedef struct wavRiff_s {
    uint32_t chunkID;
    uint32_t chunkSize;
    uint32_t format;
} wavRiff_t;

typedef struct wavProperties_s {
    uint32_t chunkID;
    uint32_t chunkSize;
    uint16_t audioFormat;
    uint16_t numChannels;
    uint32_t sampleRate;
    uint32_t byteRate;
    uint16_t blockAlign;
    uint16_t bitsPerSample;
} wavProperties_t;
/* variables hold file, state of process wav file and wav file properties */    
File root;
headerState_t state = HEADER_RIFF;
wavProperties_t wavProps;

//i2s configuration 
int i2s_num = 0; // i2s port number
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_t)(I2S_COMM_FORMAT_I2S | I2S_COMM_FORMAT_I2S_MSB),
     .intr_alloc_flags = ESP_INTR_FLAG_LEVEL1, // high interrupt priority
     .dma_buf_count = 8,
     .dma_buf_len = 64,   //Interrupt level 1
     .use_apll = false
    };
    
i2s_pin_config_t pin_config = {
    .bck_io_num = 26, //this is BCK pin
    .ws_io_num = 25, // this is LRCK pin
    .data_out_num = 22, // this is DATA output pin
    .data_in_num = -1   //Not used
};
//
void debug(uint8_t *buf, int len){
  for(int i=0;i<len;i++){
    Serial.print(buf[i], HEX);
    Serial.print("\t");
  }
  Serial.println();
}
/* write sample data to I2S */
int i2s_write_sample_nb(uint32_t sample){
  return i2s_write_bytes((i2s_port_t)i2s_num, (const char *)&sample, sizeof(uint32_t), 100);
}
/* read 4 bytes of data from wav file */
int read4bytes(File file, uint32_t *chunkId){
  int n = file.read((uint8_t *)chunkId, sizeof(uint32_t));
  return n;
}

/* these are function to process wav file */
int readRiff(File file, wavRiff_t *wavRiff){
  int n = file.read((uint8_t *)wavRiff, sizeof(wavRiff_t));
  return n;
}
int readProps(File file, wavProperties_t *wavProps){
  int n = file.read((uint8_t *)wavProps, sizeof(wavProperties_t));
  return n;
}

void setup()
{
  Serial.begin(115200);
  Serial.print("Initializing SD card...");
  if (!SD.begin(33)) {
    Serial.println("initialization failed!");
    return;
  }
  Serial.println("initialization done.");
  delay(1000);
  
  /* open wav file and process it */
  root = SD.open("/LR_STEREO.WAV");  //"LR_STEREO.WAV" or "LR_MONO.WAV"
  if (root) {    
    int c = 0;
    int n;
    while (root.available()) {
      switch(state){
        case HEADER_RIFF:
        wavRiff_t wavRiff;
        n = readRiff(root, &wavRiff);
        if(n == sizeof(wavRiff_t)){
          if(wavRiff.chunkID == CCCC('R', 'I', 'F', 'F') && wavRiff.format == CCCC('W', 'A', 'V', 'E')){
            state = HEADER_FMT;
            Serial.println("HEADER_RIFF");
          }
        }
        break;
        case HEADER_FMT:
        n = readProps(root, &wavProps);
        if(n == sizeof(wavProperties_t)){
          state = HEADER_DATA;
        }
        break;
        case HEADER_DATA:
        uint32_t chunkId, chunkSize;
        n = read4bytes(root, &chunkId);
        if(n == 4){
          if(chunkId == CCCC('d', 'a', 't', 'a')){
            Serial.println("HEADER_DATA");
          }
        }
        n = read4bytes(root, &chunkSize);
        if(n == 4){
          Serial.println("prepare data");
          state = DATA;
        }
        //initialize i2s with configurations above
        i2s_driver_install((i2s_port_t)i2s_num, &i2s_config, 0, NULL);
        i2s_set_pin((i2s_port_t)i2s_num, &pin_config);
        break; 
        /* after processing wav file, it is time to process music data */
        case DATA:
        uint32_t data; 
        n = read4bytes(root, &data);
        i2s_write_sample_nb(data); 
        break;
      }
    }
    root.close();
  } else {
    Serial.println("error opening file");
  }
  i2s_driver_uninstall((i2s_port_t)i2s_num); //stop & destroy i2s driver 
  Serial.println("done!");
}

void loop()
{
}

I've changing some of the configuration parameters, such as the sample rate, bits per sample, and channel format, but regardless of what I change, I can't get the stereo wav file to play at 1x speed without a lot of noise in the background. If you need the audio files, I can try and attach them if needed, but does anyone know what the problem might be? I've been trying to get this working for a few days now with no luck.

[awwende]

I think I found what the problem is, in the i2s_config there are settings for the dma buffer count and length. If I adjust the values, I can reduce the noise, or if they're too large, they seem to introduce echoing/reverb. I'm not sure what the correct values are though.

In this example (https://github.com/espressif/esp-idf/bl ... ple_main.c) they somewhat mention how the values were selected, but my audio files aren't at a fixed frequency so I'm still not sure what values I should use for my 16-bit, 44.1kHz sample rate stereo audio file.