using 24bit-96khz output DAC

[josef2600]

for simplicity, for starting point, i am using ESP32-A1S board. it has Es8388 DAC.
what i want is to use the maximum bit rate and frequency of the DAC to get the best quality.
the Bluetooth A2DP, as i can understand and for simplicity, works with 44100khz and 16bit. CD quality apparently!
but this DAC can output 24bit and 96khz.
so i want to convert 16bit 44100khz to 24bit and 96khz.
for this purpose, i found 2 functions: convert.begin(16, 32); and resample.begin(from_info, to_info.sample_rate); .
my problem is that neither works! in both situations the sound is destroyed! it works with 44100khz and 16bit. but not with 96khz or 24bit.
i tested all these lines:
AudioInfo to_info(44100, 2, 24);
and
AudioInfo to_info(96000, 2, 24);
and
AudioInfo to_info(96000, 2, 16);
but non of them worked.

#define AUDIOKIT_BOARD  7
//#define AI_THINKER_ES8388_VOLUME_HACK  0
#define DRIVER_DEFAULT_VOLUME 90
#define ES8388_DEFAULT_INPUT_GAIN 25
#define ES8388_PA_LINE 1

#include "AudioTools.h"
#include "BluetoothA2DPSink.h"
//#include "AudioTools/AudioLibs/A2DPStream.h" // install https://github.com/pschatzmann/ESP32-A2DP
#include "AudioTools/AudioLibs/AudioBoardStream.h" // install https://github.com/pschatzmann/arduino-audio-driver

AudioBoardStream kit(AudioKitEs8388V2);  //AudioBoard.h  static AudioBoard AudioKitEs8388V2{AudioDriverES8388, PinsAudioKitEs8388v2};

AudioInfo from_info(44100, 2, 16);
AudioInfo to_info(44100, 2, 24);
//I2SStream out;
ResampleStream resample(kit);
BluetoothA2DPSink a2dp_sink(resample);

void setup()
{
  Serial.begin(115200);
  delay(500);
  AudioLogger::instance().begin(Serial, AudioLogger::Info);
  delay(100);
  // setup i2s with 48000
  auto cfg = kit.defaultConfig();
  cfg.sd_active = false;
  cfg.auto_clear = true;
  cfg.copyFrom(to_info);
  //kit.setVolume(0.9);
  kit.begin(cfg);
  // setup resample from 44100 to 48000
  resample.begin(from_info, to_info.sample_rate);
  // start a2dp
  a2dp_sink.start("bt_MyMusic");
  delay(10);
}

void loop()
{
  delay(1000); // do nothing
}

and the convert.begin(16, 32); sample doesn't work either.
but the funny thing is in the logging:
it say's both in and out sample rates are the same. why?

Warning: Pin '23' not set up because of conflict
Warning: Pin '18' not set up because of conflict
[I] AudioTypes.h : 128 - in: sample_rate: 44100 / channels: 2 / bits_per_sample: 24
[I] AudioTypes.h : 128 - out: sample_rate: 44100 / channels: 2 / bits_per_sample: 24
[I] AudioTypes.h : 128 -  sample_rate: 44100 / channels: 2 / bits_per_sample: 24
[I] I2SConfigESP32.h : 80 - rx/tx mode: TX_MODE
[I] I2SConfigESP32.h : 81 - port_no: 0
[I] I2SConfigESP32.h : 82 - is_master: Master
[I] I2SConfigESP32.h : 83 - sample rate: 44100
[I] I2SConfigESP32.h : 84 - bits per sample: 24
[I] I2SConfigESP32.h : 85 - number of channels: 2
[I] I2SConfigESP32.h : 86 - signal_type: Digital
[I] I2SConfigESP32.h : 88 - i2s_format: I2S_STD_FORMAT
[I] I2SConfigESP32.h : 90 - auto_clear: true
[I] I2SConfigESP32.h : 92 - use_apll: true
[I] I2SConfigESP32.h : 97 - buffer_count:6
[I] I2SConfigESP32.h : 98 - buffer_size:512
[I] I2SConfigESP32.h : 101 - pin_mck: 0
[I] I2SConfigESP32.h : 103 - pin_bck: 5
[I] I2SConfigESP32.h : 105 - pin_ws: 25
[I] I2SConfigESP32.h : 107 - pin_data: 26
[I] ResampleStream.h : 64 - begin step_size: 1.000000
[I] ResampleStream.h : 145 - setStepSize: 1.000000
[I] ResampleStream.h : 145 - setStepSize: 1.000000
[I] ResampleStream.h : 129 - -> ResampleStream:
[I] AudioTypes.h : 128 - out: sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] ResampleStream.h : 64 - begin step_size: 1.000000
[I] ResampleStream.h : 145 - setStepSize: 1.000000
[I] ResampleStream.h : 145 - setStepSize: 1.000000
[I] ResampleStream.h : 129 - -> ResampleStream:
[I] AudioTypes.h : 128 - out: sample_rate: 44100 / channels: 2 / bits_per_sample: 16

and for the NumberFormatConverterStream convert(kit);

#include "AudioTools.h"
#include "BluetoothA2DPSink.h"
#define AUDIOKIT_BOARD  7
#define ES8388_PA_LINE 1

#include "AudioTools.h"
#include "BluetoothA2DPSink.h"
#include "AudioTools/AudioLibs/AudioBoardStream.h" // install https://github.com/pschatzmann/arduino-audio-driver
AudioBoardStream kit(AudioKitEs8388V2); 
AudioInfo info(44100, 2, 16);
NumberFormatConverterStream convert(kit);
BluetoothA2DPSink a2dp_sink(convert);

void setup()
{
  Serial.begin(115200);
  delay(500);
  AudioLogger::instance().begin(Serial, AudioLogger::Info);
  delay(100);
  // Configure i2s to use 32 bits
  auto cfg = kit.defaultConfig();
  cfg.copyFrom(info);
  cfg.sd_active = false;
  cfg.auto_clear = true;
  
  kit.begin(cfg);
  kit.setVolume(0.9);
  // Convert from 16 to 24 bits
  convert.begin(16, 24);
  // start a2dp
  a2dp_sink.start("bt_AudioKit");
  delay(10);
}

void loop()
{
  delay(1000); // do nothing
}

with this code, i only get nois. hisssssssssssssssssssssssssss. no audio at all when playing. if i stop playing, the hissss goes away.

Warning: Pin '23' not set up because of conflict
Warning: Pin '18' not set up because of conflict
[I] AudioTypes.h : 128 - out: sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] AudioTypes.h : 128 -  sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] I2SConfigESP32.h : 80 - rx/tx mode: TX_MODE
[I] I2SConfigESP32.h : 81 - port_no: 0
[I] I2SConfigESP32.h : 82 - is_master: Master
[I] I2SConfigESP32.h : 83 - sample rate: 44100
[I] I2SConfigESP32.h : 84 - bits per sample: 16
[I] I2SConfigESP32.h : 85 - number of channels: 2
[I] I2SConfigESP32.h : 86 - signal_type: Digital
[I] I2SConfigESP32.h : 88 - i2s_format: I2S_STD_FORMAT
[I] I2SConfigESP32.h : 90 - auto_clear: true
[I] I2SConfigESP32.h : 92 - use_apll: true
[I] I2SConfigESP32.h : 97 - buffer_count:6
[I] I2SConfigESP32.h : 98 - buffer_size:512
[I] I2SConfigESP32.h : 101 - pin_mck: 0
[I] I2SConfigESP32.h : 103 - pin_bck: 5
[I] I2SConfigESP32.h : 105 - pin_ws: 25
[I] I2SConfigESP32.h : 107 - pin_data: 26
[I] AudioStreamsConverter.h : 490 - begin 16 -> 24 bits
[I] AudioStreamsConverter.h : 490 - begin 16 -> 24 bits

and if i change to AudioInfo info(44100, 2, 24); , no output at all and i get this:

ASSERT_WARN(1 0), in lc_task.c at line 9303[I] AudioStreamsConverter.h : 490 - begin 16 -> 24 bits
ASSERT_WARN(1 36), in lc_task.c at line 2300

a little help would be wonderful. thanks.

[josef2600]

then i did this:

#include "AudioTools.h"
#include "BluetoothA2DPSink.h"
#define AUDIOKIT_BOARD  7
#define ES8388_PA_LINE 1
#include "AudioTools.h"
#include "BluetoothA2DPSink.h"
#include "AudioTools/AudioLibs/AudioBoardStream.h" // install https://github.com/pschatzmann/arduino-audio-driver

AudioBoardStream kit(AudioKitEs8388V2); 

AudioInfo info(44100, 2, 24);
BluetoothA2DPSink a2dp_sink(kit);

// callback
void get_audio_data(const uint8_t *data, uint32_t len)
{
  // data is provided as array of int16
  uint32_t sample_count = len / sizeof(uint16_t);
  uint16_t* data16 = (uint16_t*) data;
  for (int j = 0; j < sample_count; j++)
  {
    // convert to 24 bit data
    uint32_t sample32 = static_cast<uint32_t>(data16[j]) << 8;
    // add your special logic here
    // output 24 bits to DAC
    //kit.write((uint8_t*)&sample32, sizeof(sample32));
    kit.write((uint8_t*)&sample32, sizeof(sample32));
  }
}

void setup()
{
  Serial.begin(115200);
  delay(500);
  AudioLogger::instance().begin(Serial, AudioLogger::Info);
  delay(100);
  // Configure i2s to use 32 bits
  auto cfg = kit.defaultConfig();
  cfg.copyFrom(info);
  cfg.sd_active = false;
  cfg.auto_clear = true;
  
  kit.begin(cfg);
  kit.setVolume(0.9);
  
  // data callback with no output
  a2dp_sink.set_stream_reader(get_audio_data, false);
  // start a2dp
  a2dp_sink.start("bt_AudioKit");
  delay(10);
}

void loop()
{
  delay(1); // do nothing
}

it plays audio in slow motion!
and the info says:

Warning: Pin '23' not set up because of conflict
Warning: Pin '18' not set up because of conflict
[I] AudioTypes.h : 128 - in: sample_rate: 44100 / channels: 2 / bits_per_sample: 24
[I] AudioTypes.h : 128 - out: sample_rate: 44100 / channels: 2 / bits_per_sample: 24
[I] AudioTypes.h : 128 -  sample_rate: 44100 / channels: 2 / bits_per_sample: 24
[I] I2SConfigESP32.h : 80 - rx/tx mode: TX_MODE
[I] I2SConfigESP32.h : 81 - port_no: 0
[I] I2SConfigESP32.h : 82 - is_master: Master
[I] I2SConfigESP32.h : 83 - sample rate: 44100
[I] I2SConfigESP32.h : 84 - bits per sample: 24
[I] I2SConfigESP32.h : 85 - number of channels: 2
[I] I2SConfigESP32.h : 86 - signal_type: Digital
[I] I2SConfigESP32.h : 88 - i2s_format: I2S_STD_FORMAT
[I] I2SConfigESP32.h : 90 - auto_clear: true
[I] I2SConfigESP32.h : 92 - use_apll: true
[I] I2SConfigESP32.h : 97 - buffer_count:6
[I] I2SConfigESP32.h : 98 - buffer_size:512
[I] I2SConfigESP32.h : 101 - pin_mck: 0
[I] I2SConfigESP32.h : 103 - pin_bck: 5
[I] I2SConfigESP32.h : 105 - pin_ws: 25
[I] I2SConfigESP32.h : 107 - pin_data: 26
[I] I2SCodecStream.h : 93 - virtual void audio_tools::I2SCodecStream::setAudioInfo(audio_tools::AudioInfo)
[I] AudioTypes.h : 128 - in: sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] AudioTypes.h : 128 - out: sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] I2SStream.h : 93 - virtual void audio_tools::I2SStream::setAudioInfo(audio_tools::AudioInfo)
[I] AudioTypes.h : 128 - in: sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] AudioTypes.h : 128 - out: sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] I2SStream.h : 102 - restarting i2s
[I] AudioTypes.h : 128 - I2SStream sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] AudioTypes.h : 128 -  sample_rate: 44100 / channels: 2 / bits_per_sample: 16
[I] I2SConfigESP32.h : 80 - rx/tx mode: TX_MODE
[I] I2SConfigESP32.h : 81 - port_no: 0
[I] I2SConfigESP32.h : 82 - is_master: Master
[I] I2SConfigESP32.h : 83 - sample rate: 44100
[I] I2SConfigESP32.h : 84 - bits per sample: 16
[I] I2SConfigESP32.h : 85 - number of channels: 2
[I] I2SConfigESP32.h : 86 - signal_type: Digital
[I] I2SConfigESP32.h : 88 - i2s_format: I2S_STD_FORMAT
[I] I2SConfigESP32.h : 90 - auto_clear: true
[I] I2SConfigESP32.h : 92 - use_apll: true
[I] I2SConfigESP32.h : 97 - buffer_count:6
[I] I2SConfigESP32.h : 98 - buffer_size:512
[I] I2SConfigESP32.h : 101 - pin_mck: 0
[I] I2SConfigESP32.h : 103 - pin_bck: 5
[I] I2SConfigESP32.h : 105 - pin_ws: 25
[I] I2SConfigESP32.h : 107 - pin_data: 26

if i do this kit.write((uint8_t*)&sample32,3); , i only get hisssss.

[josef2600]

and with AudioInfo info(44100, 2, 32); it gets even slower!

[josef2600]

if i do this AudioInfo info(44100, 2, 16); , i only get hissss. but in debug, it doesn't reset the i2s.

[pschatzmann]

Why don't you just use a FormatConverterStream and let it do it's work ?

By the way your assumption that this would increase the quality is just plain wrong!

[josef2600]

thank you very much!
why would you say your assumption that this would increase the quality is just plain wrong! ? you didn't specified why?
and i did say the sample for resample.begin(from_info, to_info.sample_rate); didnt work.
but i was looking trough the es8388 driver. i think i am going the wrong way about it. it should be set in the driver to use maximum resolution, not in the audio codec.
but i am going to plain I2S mode to see whats what.

[pschatzmann]

The audio is transmitted with 16 bits at 44100 samples per second so that's the quality you get.
You can convert it to 24 or 32 bits: but you don't gain any information from this operation; The same holds true with resampling.

The AudioBoardStream configures both: I2S and the es8388 Codec.

[josef2600]

thank you. but you are wrong about resampling. we have 2 kind of work in DAC. first part is the input data. in that part, you are correct. but there is a second part too. the output part. the sampling rate and bit rate of the output for DAC. it is both relevant and irrelevant to the input. if the input is faster or has a higher bit rate than the output , the audio gets bad. but if the output for the DAC is faster, you get a lot better quality. this one is physics, not code. for example, for capacitor rectifier, with the same capacitor , the output gets lower noise if the frequency goes higher. simple electronic physic.

[pschatzmann]

It would be interresting if you would be able to measure this.

My point was that you will not get a higher frequency response by upsampling. I am also using a imple linear interpolation to determine the "missing" samples which might have it's drawbacks.

[pschatzmann]

Do you mean the mclk pin ?

[josef2600]

no. the ES8388 needs a clock trough that pin. it is over 12mhz.

[josef2600]

i need to change that 12mhz clock

[pschatzmann]

That's part of the I2S definition in the AudioTools: mclk_multiple.
By default the ES8388 is configured as slave and I am using the defaut settings of the mclk

[josef2600]

ok. so you dont give it a number. i should search for it. thanks.

[pschatzmann]

There are some comments about this in the IDF documentation.

[josef2600]

so, i measured the frequency by oscilloscope. it is around 11.2 mhz. it means it should be 11.2896 MHz. base on the chart, these are the only sample rates it could work with:
8.0182 kHz
11.025 kHz
22.05 kHz
44.1 kHz
88.2 kHz
both ADC and DAC.
so, no 96khz with that. and since the clocks for DAC are drives from MCLK (main clock) , for getting 44.1 kHz input to work, i have to use the 11.2896 MHz clock (the default). bad chip!!!
now let me check if i can change the DAC clock to 88.2 kHz and get the input data for 44.1 kHz. and then bit rates.

[josef2600]

i have not tested it, but for reminder i put it here. i may change it.
i added a function for ADC clock (set ADC Sample Rate (Frequency)) . i have not tested it yet or compile it. but seen a lot of mismatches in registers. if the datasheet is correct (!) and i didnt mess up, there are many errors in driver . i will upload the full version after my tests. but here it is.

/**
 * @brief Configure ES8388 I2S format
 *
 * @param mode:           set ADC Sample Rate (Frequency)
 * @param bitPerSample:   see Es8388I2sFmt
 *
 * @return
 *     - (-1) Error
 *     - (0)  Success
 */
error_t es8388_config_ADC_Sample_Rate(uint32_t Sample_Rate)
{
  AD_LOGD(LOG_METHOD);
  error_t res = RESULT_OK;
  uint8_t reg = 0;
  if (mode == CODEC_MODE_ENCODE || mode == CODEC_MODE_BOTH)
  {
    reg = 0b000010;
    if (Sample_Rate == 8018)  reg = 0b01001;// (MCLK/1408)
    else if (Sample_Rate == 11025)  reg = 0b000111;// (MCLK/1024)
    else if (Sample_Rate == 22050)  reg = 0b000100;// (MCLK/512)
    else if (Sample_Rate == 44100)  reg = 0b000010;//  (MCLK/256)
    else if (Sample_Rate == 88200)  reg = 0;// (MCLK/128)
    
    res = es_write_reg(ES8388_ADDR, ES8388_ADCCONTROL5, reg);// set both adc channels to 16 bit.minimum is 16bit.
  }
  
  return res;
}

this should work if Register 8 – Master Mode Control , works with the setting it says! oh, i changed that too.
and i have to say the first option (8018) is not exactly 8018 , but 8018.2 !!
and ADC is set to 16bit only.

[pschatzmann]

I wasn't testing the ESP8388 in master mode and only used it in slave mode.
If you need some additonal logic, I suggest to extend e.g. the es8388_config_i2s() method where you get all information you need.

[josef2600]

funny, i was reading the new driver for es8388 from espressif / esp-adf .they address the chip with #define ES8388_ADDR 0x20 /*!< 0x22:CE=1;0x20:CE=0*/ . but your driver is #define ES8388_ADDR 0x10 . base on manual we have this:
The chip address must be 001000x, where x equals AD0. The RW bit indicates the slave data transfer direction.
and in another part of manual we have this:
* If PIN26 (CE) be pulled down to GND, I2C Chip ID is 0x20, otherwise 0x22.
i am very confused.
and in your driver, it has said /* ES8388 address 0x22:CE=1;0x20:CE=0 */ but then selected #define ES8388_ADDR 0x10

[pschatzmann]

Wire has different conventions then the IDF for addresses: They are shifted by 1 bit
I was wondering first why the IDF addresses did not work...

[josef2600]

we have a big problem. at least i have!
in the config for I2S we have :

  cfg.i2s.bits = BIT_LENGTH_16BITS;
  cfg.i2s.rate = RATE_44K;

but if our codec has ADC and DAC, they will config the same. but they can (and should) have different setting. to me, it is very important. can you add (or fix) this issue? for compatibility, dont change the defualt, only add 2 extra line for ADC setting. like:

  cfg.i2s.bits = BIT_LENGTH_16BITS;
  cfg.i2s.rate = RATE_44K;

  cfg.i2s.adcbits = BIT_LENGTH_16BITS;
  cfg.i2s.adcrate = RATE_44K;

this way no one has to change anything, but if they want, they can change it.
and for simplicity, for the code, i didn't want to change the base code for the driver, so i add the code like this:

error_t es8388_config_ADC_Sample_Rate(uint32_t Sample_Rate);
error_t es8388_config_DAC_Sample_Rate(uint32_t Sample_Rate);

and i didn't remove or modify the names of the main code:
error_t es8388_i2s_config_clock(es_i2s_clock_t cfg)
but originally, this function changes the frequency for both ADC and DAC. this function doesn't have the input mode (DAC-ADC).
lets say i want the high quality for DAC but low quality and fast speed for ADC. so i don't clog the esp32.

but what way do you want it? should i change the error_t es8388_i2s_config_clock(es_i2s_clock_t cfg) so it would be like error_t es8388_i2s_config_clock(int mode, es_i2s_clock_t cfg) ? since the drivers are not written in c++ , i don't think it would be wise to do them both:

error_t es8388_i2s_config_clock(es_i2s_clock_t cfg);
error_t es8388_i2s_config_clock(int mode, es_i2s_clock_t cfg);

or do nothing to error_t es8388_i2s_config_clock(es_i2s_clock_t cfg) and it does what it does, but have

error_t es8388_config_ADC_Sample_Rate(uint32_t Sample_Rate);
error_t es8388_config_DAC_Sample_Rate(uint32_t Sample_Rate);

to do those things separately?

what is your opinion?

[pschatzmann]

I am afraid I don't quite understand: if you use the ES8388 as DAC and ADC at the same time, you will operate I2S in douplex mode, so it is critical that they use the same rates!

In fact this is how the AudioKit, LyraT or all other audio boards are working.

So having different sample rates does not make any sens to me.

[josef2600]

oh, but they do.
if a DAC has ADC too, it has different options for them. but there is always 1 pin for ADC output too. if we check the pin 8 of es8388, it is called ASDOUT. this pin is the data out for ADC only. the DAC and ADC share the clock pin and left/right pin. 1 pin is input data for DAC and 2 pins for I2C.
that is ok. in my driver, i rewritten almost all of it. it had so many problems.
so far, i haven't changed breaking code. so it will work with your code without changes. but it has to be changed. as i said before, it is important for calculation of the data. usually users don't need high quality ADC, but they might, and if they do, they can change it.
i just finished my draft. i have to test it. if it worked i will share it with you, then i will explain how it works. i hope we can fix the problem with the bit rate later. as it was the problem i started with!

[josef2600]

so, i finished it for test. not everything is good. volume should be fixed.

es8388-worked.zip

this is a simple code i tested it:

#define AUDIOKIT_BOARD  7

#include "AudioTools.h"
#include "AudioTools/AudioLibs/A2DPStream.h" // install https://github.com/pschatzmann/ESP32-A2DP
#include "AudioTools/AudioLibs/AudioBoardStream.h" // install https://github.com/pschatzmann/arduino-audio-driver

BluetoothA2DPSink a2dp_sink;
AudioBoardStream kit(AudioKitEs8388V2);  //AudioBoard.h  static AudioBoard AudioKitEs8388V2{AudioDriverES8388, PinsAudioKitEs8388v2};

// Write data to AudioKit in callback
void read_data_stream(const uint8_t *dataaa, uint32_t lengthh)
{
  kit.write(dataaa, lengthh);
}

void setup()
{
  Serial.begin(115200);
  delay(500);
  AudioLogger::instance().begin(Serial, AudioLogger::Info);
  delay(100);
  
  // setup output
  auto cfg = kit.defaultConfig(TX_MODE);
  cfg.sd_active = false;
  kit.begin(cfg);
  delay(10);
  a2dp_sink.set_stream_reader(read_data_stream, false);
  
  a2dp_sink.start("AudioKit");
  delay(10);
}

void loop()
{
  delay(10);
}

it should work with other codes too. i have not tested the mic or line in yet.
but what do you think? does it sound better or not?

[josef2600]

es8388.zip
this one is good. the output is good, the volume works fine. the microphone is not tested yet.
but i have added this function to it:
es8388_extra__voice_volume(uint8_t volumee);
it is for extra volume. on default, it is set on 0DB. but it goes to 4.5DB if you give it 5. it was messed up before and was in the volume hacks, not a good implementation.
the es8388_get_voice_volume(int *volume) works good.

[josef2600]

in es8388_init(codec_config_t *cfg, i2c_bus_handle_t handle) and es8388_config_input_device(es8388_input_device_t inputt) i have added this:

  #ifdef ADC_INPUT_DSR1
  else if (inputt == ADC_INPUT_DSR1)  //LINPUT1-RINPUT1 differential INPUT
  {
    res |= es_write_reg(ES8388_ADDR, ES8388_ADCCONTROL2, esChip_ADC_Control_2_DSR1);
    AD_LOGI("ADC_Control_2_DSR1: 0x%x",inputt);
  }
  #endif
  #ifdef ADC_INPUT_DSR2
  else if (inputt == ADC_INPUT_DSR2)  //LINPUT2-RINPUT2 differential INPUT
  {
    res |= es_write_reg(ES8388_ADDR, ES8388_ADCCONTROL2, esChip_ADC_Control_2_DSR2);
    AD_LOGI("ADC_Control_2_DSR2: 0x%x",inputt);
  }
  #endif

since you didn't define ADC_INPUT_DSR1 and ADC_INPUT_DSR2, i put them in #ifdef. it is for selecting differential inputs. for line 1 and line 2 inputs. if you like, you can add them too.
es8388.zip

[josef2600]

and we have different sample rates for ADC and DAC , if you want to use them:

es8388_config_ADC_Sample_Rate(uint32_t Sample_Rate)
es8388_config_DAC_Sample_Rate(uint32_t Sample_Rate)

we only have these 5 sample rate for the default frequency of I2S. default is 44100 Hz for both.
there are few problem with using es8388_i2s_config_clock(es_i2s_clock_t cfg). since they are not clear to use what? the ADC and DAC should be different. and i have to check to see what it sends there. so, i only put the default 44100khz there.
but in es8388_init(codec_config_t *cfg, i2c_bus_handle_t handle) it does set them both.

if (cfg->i2s.bits == 24)
....
if (cfg->i2s.rate == 8018)  tmpp = 0b01001;// (MCLK/1408)
...

[josef2600]

es8388.zip
added a funny function! you can choose to send from input directly to output!

/**
 * @param es8388_mixer_to_output
 * 
 * if you need to send the input to output directly
 * or choose what to send to output, then choose an input
 * 0 = DAC to output
 * 1 = LIN1 + RIN1
 * 2 = LIN2 + RIN2
 * 3 = left + right ADC P input (after mic amplifier)
 * 4 = left + right ADC N input (after mic amplifier)
 * @return
 *     - (-1) Parameter error
 *     - (0)   Success
 */
error_t es8388_mixer_to_output(uint8_t channell)

and fixed few registers options.

do you want me to continue here?

[josef2600]

just for one example, how bad the original driver is:
in the user manual, it says for Power Down Sequence (To Standby Mode) (and in original driver):

first step:
Mute ADC and DAC
Reg 0x0F = 0x34 (ADC Mute)
Reg 0x19 = 0x36 (DAC Mute)

you know the funny thing?
in the datasheet, in page 19 we have:

6.2.7 Register 15 – ADC Control 7, Default 0010 0000
ADCMute
0 – normal (default) 
1 – mute ADC digital output

but 0x34 does not mute it!.
i don't have time for all of their mistakes and wrong registers and works.
but my new driver will (and has) fixed almost all of them.

[pschatzmann]

Feel free to submbit a pull request after you have tested that everything is still working.

I guess the simplest test sketch to use would be this one.

I was quickly double checking your last version and it did not produce any output with this test: so something is definitly not good.

It would be definitly a good thing if the ESP32 would be supported also in slave mode

[josef2600]

thanks. that is why i asked. i don't have every board and cant test many codes.
but to this one, mine is the same module, but without the amp and keys ... . and i simply only tested the Bluetooth.
please let me check them. but, the keys and selectors and ... are not related to es8388 driver! but there is only check for headphone jack test.
after i tested few things, i will give you a new version. and i must say, since i am differ the ADC and DAC in all settings, it is very complicated to implant it.
also there is the output selector too. i have separated them too. but no matter. i will fix them, or i will do something!

[josef2600]

we have a problem! for input selecting, there are 2 main inputs:
(LIN1+ RIN1) and (LIN2+ RIN2). now, these are left input and right input 1&2. in module level. but, in your code we have
input_device = ADC_INPUT_NONE or ADC_INPUT_LINE1 or ADC_INPUT_LINE2 or ADC_INPUT_ALL .now problem is this:
the inputs are (LIN1+ RIN1) and (LIN2+ RIN2). but in the driver level (inside es8388) there are many selections:
(LIN1+ RIN1) with no mic amp or (LIN1+ RIN1) + mic amp or (LIN2+ RIN2) with no mic amp or (LIN2+ RIN2) + mic amp or
(LIN1+ LIN2) in deferential mode with no mic amp or (RIN2+ RIN2) in deferential mode + mic amp or
(LIN1+ LIN2) in deferential mode with no mic amp AND (RIN2+ RIN2) in deferential mode with no mic amp or
(LIN1+ LIN2) in deferential mode + mic amp AND (RIN2+ RIN2) in deferential mode + mic amp or mix of them. you get the point!
deferential mode means like this:
LIN1 is input positive and LIN2 is input negative.
now you see my dilemma? how am i suppose to config the input? need help!

[josef2600]

just to be more clear, there are no mic input in es8388. you can check the datasheet in pins. don't get confused. in the module pins, it says it has mic inputs, it is because of a circuit it has for driving a microphone. it is not a special input for es8388. and base on the schematic of esp32-a1s_v2.3 we have this:

MIC2N to RIN2
MIC2P to LIN2
MIC1P to RIN1
MIC1N to LIN1

[pschatzmann]

This is also the reason why the neutral names ADC_INPUT_LINE1 and ADC_INPUT_LINE1 are used in tze API.
Just take the current working implentation on a LyraT Board as baseline.

As far as I know the input is messed up on the hardware side for the AudioKit and I don't see how we can fix this. The output side is working correctly there.

[josef2600]

can i change class CodecConfig : public codec_config_t and add some to it? it will have zero effect to you codes. i need to add some modes to it, but i will not change any of current config. but since it is in your multiple libraries, you have to do it.
like i tried to do this:

namespace audio_driver {
...

typedef enum
{
  ADC_INPUT_NONE = 0x00, /*!< no input */
  ADC_INPUT_LINE1 = 0,       /*!< mic input from adc 1 */
  ADC_INPUT_LINE2 = 0,       /*!< mic input from adc 2 */
  ADC_INPUT_LINE3 = 0,       /*!< mic input from adc 3 */
  ADC_INPUT_ALL = 0,         /*!< mic input from all adc */
  ADC_INPUT_DIFFERENCE = 0,  /*!< mic input to adc difference channel */
} ADC_input_dve_t;

typedef enum
{
  // with the speed of 11.2896 MHz for I2S, we dont have 16 khz.only these speeds:
  RATE_8K  = 8, /*!< set to  8018 samples per second */
  RATE_11K = 11,    /*!< set to 11.025k samples per second */
  RATE_22K = 22,    /*!< set to 22.050k samples per second */
  RATE_44K = 44,    /*!< set to 44.1k samples per second */
  RATE_88K = 88,    /*!< set to 88.2k samples per second */
} ADC_dve_rate;

typedef enum
{
  BIT_LENGTH_MIN = 16,
  BIT_LENGTH_16BITS = 16,
  BIT_LENGTH_18BITS = 18,
  BIT_LENGTH_20BITS = 20,
  BIT_LENGTH_24BITS = 24,
  BIT_LENGTH_32BITS = 32,
  BIT_LENGTH_MAX = 32,
} ADC_dve_bits;

and i got a hundred errors from other libs, like arduino-audio-tools-main.
so i have a simple question: can you change the audio_driver config to fit the new driver or not? in both libraries (or every where they are).
if yes, i will give you extra codes for it.
thank you.

[pschatzmann]

No the generic API should not change and should be consistent across all drivers!
So the changes should be only in the driver.

In addition I don't see any added value:

• ADC_dve_bits is already covered by sample_bits_t
• ADC_dve_rate is a subset of samplerate_t
• ADC_input_dve_t is already covered by input_device_t

[josef2600]

ok!

[josef2600]

i have to say, some configs are not possible, like (ADC_INPUT_ALL == cfg->input_device)// this is not possible!
and as i have told you, so many wrong and missing configs.

[pschatzmann]

No worries: this is currently not working as well.
Maybe I should add some warnings for the cases that are not supported....

[josef2600]

1. do you use error_t es8388_start(codec_mode_t mode) in your side or not? because if you are not using it, it is good. and all relevant settings are set in error_t es8388_init(codec_config_t *cfg, i2c_bus_handle_t handle) .
2. i have disabled error_t es8388_deinit(void) function in my code. because the necessary functions only available in error_t es8388_init(codec_config_t *cfg, i2c_bus_handle_t handle) . this is one of the wrong settings in the driver. the es8388_deinit and error_t es8388_stop(codec_mode_t mode) and similar, set some registers in es8388 that resets all settings in es8388. all bit rates and similar are lost. for that, you have to call error_t es8388_init(codec_config_t *cfg, i2c_bus_handle_t handle) again.

anyway, this is my latest driver. but i did see some odd behaviors in it. after programming the esp32, i found out i have to power off the board and Bluetooth of my phone, then power on the board then my phone. it wont work if i don't do them. the resseting the device doesnt work after program.
and please make sure you select your own board in your code! just to be sure. you have AudioBoardStream kit(AudioKitEs8388V1); i think.
please first do a simple code like this with Bluetooth mode. i have AudioKitEs8388V2.

#include "AudioTools.h"
#include "AudioTools/AudioLibs/A2DPStream.h" // install https://github.com/pschatzmann/ESP32-A2DP
#include "AudioTools/AudioLibs/AudioBoardStream.h" // install https://github.com/pschatzmann/arduino-audio-driver


BluetoothA2DPSink a2dp_sink;
//AudioBoardStream kit(AudioKitEs8388V1);  //AudioBoard.h  static AudioBoard AudioKitEs8388V2{AudioDriverES8388, PinsAudioKitEs8388v2};



// Write data to AudioKit in callback
void read_data_stream(const uint8_t *dataaa, uint32_t lengthh)
{
  kit.write(dataaa, lengthh);
}

void setup()
{
  Serial.begin(115200);
  delay(500);
  AudioLogger::instance().begin(Serial, AudioLogger::Info);
  delay(100);
  
  // setup output
  auto cfg = kit.defaultConfig(TX_MODE);
  cfg.sd_active = false;
  kit.begin(cfg);
  delay(10);
  
  // register callback
  a2dp_sink.set_stream_reader(read_data_stream, false);
  
  
  a2dp_sink.start("AudioKit");
  delay(100);
  es8388_set_voice_volume(100); //0 to 100
  es8388_extra__voice_volume(1); // 0 to 5
  delay(100);
}

void loop()
{
  delay(100);
}

es8388.zip
and tell me what happens.

[josef2600]

and if you need extra volume, you can use this:
es8388_extra__voice_volume(1); // 0 to 5
5 is the highest.

[josef2600]

fixed error for es8388_extra__voice_volume(3); . now inputs are from 0 to 4.

  if (volumee == 0)
  {
    reg = 0b011101;  // -1.5DB
    AD_LOGI("es8388_extra__voice_volume : -1.5DB");
  }
  else if(volumee == 1)
  {
    reg = 0b011110;  // 0DB
    AD_LOGI("es8388_extra__voice_volume : 0DB");
  }
  else if(volumee == 2)
  {
    reg = 0b011111;  // 1.5DB
    AD_LOGI("es8388_extra__voice_volume : 1.5DB");
    AD_LOGI("volume over 0dB may cause clipping if signal is large.");
  }
  else if(volumee == 3)
  {
    reg = 0b100000;  // 3DB
    AD_LOGI("es8388_extra__voice_volume : 3DB");
    AD_LOGI("volume over 0dB may cause clipping if signal is large.");
  }
  else if(volumee == 4)
  {
    reg = 0b100001;  // 4.5DB
    AD_LOGI("es8388_extra__voice_volume : 4.5DB");
    AD_LOGI("volume over 0dB may cause clipping if signal is large.");
  }
  else
  {
    reg = 0b011110;  // 0DB
    AD_LOGW("Warning: es8388_extra__voice_volume the volume input %d is wrong. back to 0DB", volumee);
  }

es8388.zip

[pschatzmann]

I don't qute get why you have a separate es8388_extra__voice_volume()
Why don't you just handle the the volume from 0 to 100 properly ?

[josef2600]

oh no, it is not the main volume and is not recommended to use more than 0db. it can cause clipping if signal is large. you can do it any way you like. it is separated from digital volume. but it works great.
did you test it?

[josef2600]

and since it was handled in volume hacks and it was bad, i fixed it and separate it in a function. didn't you read it? i removed volume hacks defines from it. it doesn't effect your original code, but if any one needs it, it is there.

[josef2600]

es8388.zip
its good.