Analog Audio Output Support for ESP32-S2?

[eroboticdude]

Hello, I have been experimenting with some new esp32 modules I recently got, and I want to use the newer feature of hardware USB and higher speed as compared to the esp32. I realized that this library doesn't support the DAC analog out for esp32-s2. Is this a hardware incompatibility, or is support planned?

[pschatzmann]

The implementation is basically there: you just need to figure out from which Arduino version this is available and the activate it in the config.
To my knwledge, the I2S API for the internal DAC does not work on this processor, so the new driver/dac_continuous.h API will be needed. It is my understanding that this will require ESP32 V.3.0.0

[eroboticdude]

Thanks for the reply! I'm using Arduino IDE 2.2.1 currently, and ran into the issue of the auto configuration not allowing the AnalogAudioStream to be a valid class. Is there a specific config class I need to use to change the default from I2S to the new continuous driver?

[eroboticdude]

Oh by ESP32 V3 do you mean the newest chipset? Or is it the driver/board definition that I need to update?

[pschatzmann]

As mentioned above you will need to upgrade: 2.2.1 will not work!
There are some 3.0.0 alpha releases out...

[pschatzmann]

The driver/board in the Arduino Board Manager

[eroboticdude]

Is this for Arduino IDE or the ESP32 drivers? Because the 2.2.1 Arduino IDE seems to be the latest version. Do you have an example on how to enable the new driver for the DAC? It seems that the auto configuration doesn't support this ESP32-S2 chip for output, and I'm pretty new to this libraries configuration features.

[eroboticdude]

Ok, I see now! Thanks for the help. I'll test some newer board definitions to see if it works.

[eroboticdude]

I have installed the latest alpha drivers for ESP32 boards, but I continue to get this compilation error:
Compilation error: 'AnalogAudioStream' does not name a type
What do I need to do for it to allow this class to function?

[eroboticdude]

#include <flite_arduino.h>
#include <AudioTools.h>  // https://github.com/pschatzmann/arduino-audio-tools
#include "AudioAnalog/AnalogAudioESP32V1.h"


AnalogAudioStream out; // Replace with desired class e.g. AudioKitStream, AnalogAudioStream etc.
Flite flite(out);


void setup(){
  Serial.begin(19200);
  AudioLogger::instance().begin(Serial, AudioLogger::Info);

  // start data sink
  auto cfg = out.defaultConfig();
  cfg.sample_rate = 8000;
  cfg.channels = 1;
  cfg.bits_per_sample = 16;
  out.begin(cfg);
}


// Arduino loop  
void loop() {  
        flite.say("Hello World");
  }

This is the code I used

[pschatzmann]

Did you add the USE_ANALOG in the AudioConfig.h

//-------ESP32C3, ESP32S3, ESP32S2---------

#if defined(ESP32) && defined(ESP32X)
#include "esp32-hal-log.h"

#define USE_ANALOG

[eroboticdude]

Looks like I get an exit status 1

In file included from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudioBase.h:6,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudio.h:4,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools.h:63,
                 from C:\Users\ellio\Documents\Arduino\BasicTtstest\BasicTtstest.ino:2:
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogConfigESP32V1.h:53:83: error: invalid conversion from 'int' to 'adc_channel_t' [-fpermissive]
   53 |     adc_channel_t adc_channels[3] = {ADC_CHANNEL_2, ADC_CHANNEL_3, (ADC_CHANNEL_0 | 1 << 3)};
      |                                                                    ~~~~~~~~~~~~~~~^~~~~~~~~
      |                                                                                   |
      |                                                                                   int
In file included from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools.h:56:
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/BaseConverter.h: In instantiation of 'audio_tools::ConverterAutoCenterT<T>::ConverterAutoCenterT(int) [with T = audio_tools::int24_4bytes_t]':
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/BaseConverter.h:206:73:   required from here
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/BaseConverter.h:138:20: error: conversion from 'int' to 'audio_tools::int24_4bytes_t' is ambiguous
  138 |         T offset = 0;
      |                    ^
In file included from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioConfig.h:687,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools.h:51:
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:42:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(float)'
   42 |   int24_4bytes_t(const float in) {
      |   ^~~~~~~~~~~~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:34:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(int64_t)'
   34 |   int24_4bytes_t(const int64_t in) {
      |   ^~~~~~~~~~~~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:30:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(int32_t)'
   30 |   int24_4bytes_t(const int32_t in) {
      |   ^~~~~~~~~~~~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:26:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(int16_t)'
   26 |   int24_4bytes_t(const int16_t in) {
      |   ^~~~~~~~~~~~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:22:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(void*)'
   22 |   int24_4bytes_t(void *ptr) {
      |   ^~~~~~~~~~~~~~
In file included from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTimer/AudioTimerBase.h:4,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTimer/AudioTimer.h:10,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools.h:52:
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/AudioTypes.h: In instantiation of 'static T audio_tools::NumberConverter::clip(int64_t) [with T = audio_tools::int24_4bytes_t; int64_t = long long int]':
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/AudioTypes.h:287:29:   required from 'static ToT audio_tools::NumberConverter::convert(FromT) [with FromT = short int; ToT = audio_tools::int24_4bytes_t]'
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/AudioStreamsConverter.h:292:57:   required from 'size_t audio_tools::NumberFormatConverterStreamT<TFrom, TTo>::write(const uint8_t*, size_t) [with TFrom = short int; TTo = audio_tools::int24_4bytes_t; size_t = unsigned int; uint8_t = unsigned char]'
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/AudioStreamsConverter.h:422:53:   required from here
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools/AudioTypes.h:278:25: error: conversion from 'int' to 'audio_tools::int24_4bytes_t' is ambiguous
  278 |                 return -mv;
      |                         ^~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:42:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(float)'
   42 |   int24_4bytes_t(const float in) {
      |   ^~~~~~~~~~~~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:34:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(int64_t)'
   34 |   int24_4bytes_t(const int64_t in) {
      |   ^~~~~~~~~~~~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:30:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(int32_t)'
   30 |   int24_4bytes_t(const int32_t in) {
      |   ^~~~~~~~~~~~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioBasic/Int24_4bytes_t.h:26:3: note: candidate: 'audio_tools::int24_4bytes_t::int24_4bytes_t(int16_t)'
   26 |   int24_4bytes_t(const int16_t in) {
      |   ^~~~~~~~~~~~~~

[eroboticdude]

I added this int to int24 conversion to the definition area and it removed a lot of errors:

//-------ESP32C3, ESP32S3, ESP32S2---------

#if defined(ESP32)  && defined(ESP32X)
#include "esp32-hal-log.h"

#if ESP_IDF_VERSION > ESP_IDF_VERSION_VAL(5, 0 , 0)
#  define USE_INT24_FROM_INT
#endif

#define USE_ANALOG

I now am left with only this error:

In file included from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudioBase.h:6,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudio.h:4,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools.h:63,
                 from C:\Users\ellio\Documents\Arduino\BasicTtstest\BasicTtstest.ino:2:
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogConfigESP32V1.h:53:83: error: invalid conversion from 'int' to 'adc_channel_t' [-fpermissive]
   53 |     adc_channel_t adc_channels[3] = {ADC_CHANNEL_2, ADC_CHANNEL_3, (ADC_CHANNEL_0 | 1 << 3)};
      |                                                                    ~~~~~~~~~~~~~~~^~~~~~~~~
      |                                                                                   |
      |                                                                                   int

[eroboticdude]

I removed the line that was causing the error and it compiled. It seems to work, although the TTS program seems broken and the output comes out of pin 17 and not 18. I'm going to try to set the channel count to 2.
Here's what I did to line 53 of AnalogConfigESP32V1.h:
adc_channel_t adc_channels[3] = {ADC_CHANNEL_2, ADC_CHANNEL_3, ADC_CHANNEL_0 };

I'm sure this isn't the correct approach, but it seems to work.

[eroboticdude]

After some adjustments, I can almost hear the TTS work. It is super loud and there is no clarity. Maybe there's something I can change about the gain?

[eroboticdude]

Most of the time, it sounds like static, but periodically it sounds a lot more clear. I can almost make out the test phrase.

[pschatzmann]

Thanks for your input. I have committed some corrections to add the support for the DAC and ADC automatically based on the release.
I also corrected the bug because I was providing int16_t values to an api which expects uint8_t values.
As far as the quality is concerned, I suggest to experiment with different sample rates and buffer settings in your sketch.

auto cfg = dac.defaultConfig();
cfg.buffer_size = 1024;
cfg.buffer_count = 10;
cfg.sample_rate = 8000;
dac.begin(cfg);

and to set the log level to Warning

For the volume here is the documentation in the Wiki: https://github.com/pschatzmann/arduino-audio-tools/wiki/Volume-Control

[eroboticdude]

Thanks for the help!

[eroboticdude]

It looks like there are some issues now

In file included from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudio.h:5,
                 from c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioTools.h:63,
                 from C:\Users\ellio\Documents\Arduino\BasicTtstest\BasicTtstest.ino:2:
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudioESP32V1.h: In member function 'virtual size_t audio_tools::AnalogDriverESP32V1::IO16Bit::readBytes(uint8_t*, size_t)':
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudioESP32V1.h:149:39: error: 'src' was not declared in this scope
  149 |           int16_t* data16 = (int16_t*)src;
      |                                       ^~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudioESP32V1.h:152:20: error: 'J' was not declared in this scope
  152 |             data16[J] = (static_cast<int16_t>(data[8]) - 128) * 256;
      |                    ^
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudioESP32V1.h:152:47: error: 'data' was not declared in this scope
  152 |             data16[J] = (static_cast<int16_t>(data[8]) - 128) * 256;
      |                                               ^~~~
c:\Users\ellio\Documents\Arduino\libraries\arduino-audio-tools\src/AudioAnalog/AnalogAudioESP32V1.h:152:47: note: suggested alternatives:
In file included from c:\users\ellio\appdata\local\arduino15\packages\esp32\tools\xtensa-esp32s2-elf-gcc\esp-12.2.0_20230208\xtensa-esp32s2-elf\include\c++\12.2.0\unordered_map:48,
                 from c:\users\ellio\appdata\local\arduino15\packages\esp32\tools\xtensa-esp32s2-elf-gcc\esp-12.2.0_20230208\xtensa-esp32s2-elf\include\c++\12.2.0\functional:61,
                 from C:\Users\ellio\AppData\Local\Arduino15\packages\esp32\hardware\esp32\3.0.0-alpha2\cores\esp32/HardwareSerial.h:49,
                 from C:\Users\ellio\AppData\Local\Arduino15\packages\esp32\hardware\esp32\3.0.0-alpha2\cores\esp32/Arduino.h:196,
                 from C:\Users\ellio\AppData\Local\Temp\arduino\sketches\A15499B9EC634CD0833E5002BF902F1C\sketch\BasicTtstest.ino.cpp:1:
c:\users\ellio\appdata\local\arduino15\packages\esp32\tools\xtensa-esp32s2-elf-gcc\esp-12.2.0_20230208\xtensa-esp32s2-elf\include\c++\12.2.0\bits\range_access.h:343:5: note:   'std::data'
  343 |     data(initializer_list<_Tp> __il) noexcept
      |     ^~~~
In file included from c:\users\ellio\appdata\local\arduino15\packages\esp32\tools\xtensa-esp32s2-elf-gcc\esp-12.2.0_20230208\xtensa-esp32s2-elf\include\c++\12.2.0\bits\ranges_algobase.h:39,
                 from c:\users\ellio\appdata\local\arduino15\packages\esp32\tools\xtensa-esp32s2-elf-gcc\esp-12.2.0_20230208\xtensa-esp32s2-elf\include\c++\12.2.0\bits\ranges_algo.h:35,
                 from c:\users\ellio\appdata\local\arduino15\packages\esp32\tools\xtensa-esp32s2-elf-gcc\esp-12.2.0_20230208\xtensa-esp32s2-elf\include\c++\12.2.0\algorithm:63,
                 from C:\Users\ellio\AppData\Local\Arduino15\packages\esp32\hardware\esp32\3.0.0-alpha2\cores\esp32/Arduino.h:184:
c:\users\ellio\appdata\local\arduino15\packages\esp32\tools\xtensa-esp32s2-elf-gcc\esp-12.2.0_20230208\xtensa-esp32s2-elf\include\c++\12.2.0\bits\ranges_base.h:577:43: note:   'std::ranges::__cust::data'
  577 |     inline constexpr __cust_access::_Data data{};
      |                                           ^~~~

It seems like these are for some specific areas I don't fully understand

[eroboticdude]

Is line 152 in AnalogAudioESP32V1 supposed to look like this?
data16[j] = (static_cast<int16_t>(data8[j]) - 128) * 256;
before it was just data16[j] = (static_cast<int16_t>(data[8]) - 128) * 256;

[pschatzmann]

My mistake: I forgot to commit and push my last correction
With the latest commit it should work.
Sorry for this!

[eroboticdude]

No problem :) This is quite interesting for me to see how this process goes!

[eroboticdude]

It looks like the issue with volume has been fixed and there isn't heavy distortion on the output. It must have been caused by some of those conversions?

[pschatzmann]

Yes the conversion to 8bit was critical!

[eroboticdude]

I have one more question, how do I switch the channel that the DAC uses? Currently it is outputting to pin 17, and I would like to use pin 18. Setting the volume stream to 2 channel doesn't seem to work. Is there a setting I can change, or will i have to do a channel split from 1 to 2 channels and then back to 1 channel?

[pschatzmann]

I assume that your audio is mono that you want to output to pin 18.

I just committed a correction, so that you can define the channel in the config
cfg.dac_mono_channel = DAC_CHANNEL_MASK_CH1;

[eroboticdude]

The channel mask doesn't seem to work with each individual channel, but when I set the output to 2 channels and set the mask to all it seems to work for me.

[eroboticdude]

Channel mask seems to only allow for both channels to work, not sure why

[eroboticdude]

Thanks so much for all the help!

[pschatzmann]

Here is the implemented logic:

dac_continuous_config_t cont_cfg = {
    .chan_mask =
        cfg.channels == 1 ? cfg.dac_mono_channel : DAC_CHANNEL_MASK_ALL,
    .desc_num = (uint32_t)cfg.buffer_count,
    .buf_size = (size_t) cfg.buffer_size,
    .freq_hz = (uint32_t)cfg.sample_rate,
    .offset = 0,
    .clk_src =
        cfg.use_apll
            ? DAC_DIGI_CLK_SRC_APLL
            : DAC_DIGI_CLK_SRC_DEFAULT,  // Using APLL as clock source to
                                         // get a wider frequency range
    .chan_mode = DAC_CHANNEL_MODE_ALTER,
};

So when it is mono (channels ==1) then it is taking the value from the configuration, otherwise it uses DAC_CHANNEL_MASK_ALL which sends it to both channels.