Using a M5Stack Atom Echo with Audio-Tools

[nielsnl68]

Hello,
I'm new to Audio-Tools and trying to setup an intercom (walky talky) like script using ESP-NOW using the code below. Sadly i do not get any audio from the one to the other. And i have no idea what i'm missing here.

/**
 * @file streams-i2s-i2s-2.ino
 * @brief Copy audio from I2S to I2S: We use 2 different i2s ports!
 * @author Phil Schatzmann
 * @copyright GPLv3
 */

#define DEFAULT_SAMPLE_RATE 16000
#define DEFAULT_CHANNELS 1
#define DEFAULT_BITS_PER_SAMPLE 16

#include <Arduino.h>
#include "AudioTools.h"
#include "AudioTools/Communication/ESPNowStream.h"

#include "Button2.h"

#define BUTTON_PIN 39

Button2 button;

AudioInfo info(16000, 1, 16);
ESPNowStream now;
const char* peers[] = { "ff:ff:ff:ff:ff:ff" };

I2SStream in;
StreamCopy copier_in(now, in);          // copies sound into i2s
auto config_in = in.defaultConfig(RX_MODE);

I2SStream out;
StreamCopy copier_out(out, now);                    // copies sound into i2s
auto config_out = out.defaultConfig(TX_MODE);

bool in_mode = false;

void pressed(Button2& btn) {
  out.end();
  in.begin(config_in);

  in_mode = true;
}

void released(Button2& btn) {
  in.end();
  out.begin(config_out);
  in_mode = false;
}

// Arduino Setup
void setup(void) {
  Serial.begin(115200);

  auto cfg_now = now.defaultConfig();
  cfg_now.broadcast_msg = true;
  now.begin(cfg_now);
  now.addPeers(peers);

  // start I2S in
  Serial.println("starting I2S...");

  config_in.copyFrom(info);
  config_out.port_no = 0;
  config_in.signal_type = PDM;
  config_in.pin_bck = 19;
  config_in.pin_data = 23;
  config_in.pin_ws = -1;

  // start I2S out
  config_out.copyFrom(info);
  config_out.port_no = 1;
  config_out.pin_bck = 19;
  config_out.pin_data = 22;
  config_out.pin_ws = 33;
  config_out.channel_format = audio_tools::I2SChannelSelect::Right;

  out.begin(config_out);

  button.begin(BUTTON_PIN);  //, INPUT, false
  button.setPressedHandler(pressed);
  button.setReleasedHandler(released);

  out.begin(config_out);
}

// Arduino loop - copy sound to out
void loop() {
  button.loop();
  if (in_mode) {
    copier_in.copy();
  } else {
    copier_out.copy();
  }
}

The only thing is what i hear is some clicking on the other side.

[pschatzmann]

That's way too complicated to start with.
Build your sketch in steps and test each function separately: otherwise you have no clue where the issue is....
Further suggestions can be found in this Wiki

[nielsnl68]

I have cleaned up the code above having only a button and the audio components.

[nielsnl68]

I have minimalist the code as much as i could and tried many examples, like the "streams-i2s_pdm-serial.ino" and i do not receive any audio via the serial port. Using the following configuration:

   auto cfg = i2sStream.defaultConfig(RX_MODE);
    cfg.copyFrom(info);
    cfg.signal_type = PDM;
    cfg.pin_bck = 19;
    cfg.pin_data = 23;    
    cfg.pin_data_rx = 23;    
    //cfg.use_apll = false;  
    //cfg.auto_clear = false;
    cfg.pin_ws = -1; // not used
    cfg.channel_format = audio_tools::I2SChannelSelect::Right;

    i2sStream.begin(cfg);

When i do not set the pin_blk and pin_data i see some random data but not from the mic.

[nielsnl68]

@pschatzmann i hope you could me some directions.
I know it should work because i tested it using ESPHome's audio components and it works fine there.

[pschatzmann]

Since I do not know what a M5Stack Atom Echo exactly is you will need to share some more info. Usually all necessary information is printed nicely on your board. Can you provide me with a readable picture ?

Did you manage to have the speaker working and you just have problems with the microphone now ?

[nielsnl68]

Thanks trying to help me :D

sadly i think Chatgpt was indeed hallucinating, because it would not work :(

ps. I just noticed that in your initial sketch you never call in.begin()...

The in.begin() is called in void pressed(). Just after calling the out.end()

[pschatzmann]

So what output did you get when you record a sine wave ?
What info is printed on your device ?
And what esp32 core are you using ?

[nielsnl68]

And what esp32 core are you using ?

Here is some documentation about the M5Stack Atom Echo. It is a really small unit using an ESP32-PICO-D4, a SPM1423 (PDM) mic and a NS4168 (I2S) speaker DAC.

https://docs.m5stack.com/en/atom/atomecho

I have been Looking into your code and it seems you are using the pin_bclk to set the .clk value.

Looking at the schema at the above documentation. I figure out you where half right about the ws pin. Instead of using .pin_ws = 19; it should have been .pin_bclk = 33;.

  auto config_in = in.defaultConfig(RX_MODE);
  config_in.copyFrom(info);
  config_out.port_no = 0;
  config_in.signal_type = PDM;
  config_in.pin_bck = 33;
  config_in.pin_data = 23;
  config_in.pin_ws = -1;
  in.begin(config_in);

Now i get some distorted audio using the intercom sketch. That is already a big step forward.

So what output did you get when you record a sine wave ?

I sure the output does work properly, and i did not tried to use the sine wave yet. I will try that as well to see if that is also distorted like the mic data.

What info is printed on your device ?

I'm not sure what kind of information you are looking for. Maybe you can find what you looking for on the webside above.

[pschatzmann]

I was basically looking for the pin information and device information that is printed on the device. The URL has a diagram that provides this.

If this is working then I guess you are using a very old ESP32 core version. Check in the Arduino Board Manager, select ESP32.

The actual version is 3.3.2! With that pin_ws (=lrck) must be 33; If you record a sine wave, the Arduino Serial Plotter should show you a sine wave as well...

[nielsnl68]

@pschatzmann, at the moment i have all the above working.
Only to find that the mic sound is to low for my taste.

I found the that the boost API should do the trick but adding it to my script above is something i do not get how it needs to be implemented. Using the streams-audiokit-effects-audiokit.ino sketch i did the following:

I2SStream in;
auto config_in = in.defaultConfig(RX_MODE);

AudioEffectStream effects(in); // input from i2s

SBCEncoder sbc;
EncodedAudioStream encoder(&now, &sbc);  // encode and write to ESP-now

StreamCopy copier_in(encoder, effects);  // <= this is where i'm not sure about
float BoostControl = 0.1;
Boost boost(BoostControl);

And i added in de setup()

  // setup effects
  effects.addEffect(boost);
  effects.begin(config_in);  // <= this is where i'm not sure about

  updateValues();

Could you confirm that i implemented it correct?

[nielsnl68]

I found an issue with the ESPNowStream API.
When the default_send_cb() report false then the device is locked waiting for the xSemaphore to be released. Should we not always release the xSemaphore ?