Internal ESP32 DAC

[dkm1978]

Hello.
I fork your's ESP32-audioI2S
I added internal DAC support
It works very well :)

https://www.youtube.com/watch?v=RMBcPz6vkwY

[schreibfaul1]

Hello kdm1978, good video! In many cases the 8-bit resolution of the internal DAC may be sufficient. Internally, the decoders work with a 16-bit resolution. The sound quality can be improved with an external DAC (for example PCM5102).

[dkm1978]

Thank you for your answer.
I know that the built-in converters are 8 bit but for me it is a sufficient resolution :)

[morgan-wild]

The sound you have seems great for a 8-bit DAC!
I also tried withour an external DAC and it was horrible!

[mettyw]

Hi,

I am trying to use this lib with the internal DAC of an M5Stack Grey. At least for me, the changes in m_i2s_config aren't sufficient, I hear sound but it is extremely distorted. I could fix it by altering the function Audio::Gain() as well and add something like
v[LEFTCHANNEL] = v[LEFTCHANNEL] + 0x8000;
v[RIGHTCHANNEL] = v[RIGHTCHANNEL] + 0x8000;
similar to what has been done here: https://github.com/earlephilhower/ESP8266Audio/blob/6ecb6afe3059375c19377cd94a2c99f83c64e3c8/src/AudioOutputI2S.cpp#L238 do you think it is possible to extend your code accordingly?

My understanding of ESP32, C and bit-pushing is a bit rusty, so sorry for not providing a clearer fix directly.

[mettyw]

Ah I forgot to mention that I also had to change the call to setPinout(). Actually, now I realize that all my changes are also there in @dkm1978 's original fork:

• https://github.com/dkm1978/ESP32-audioI2S/blob/5f3e30d0b3c61c48b81506545d665bbe4eb7936b/src/Audio.cpp#L143
• https://github.com/dkm1978/ESP32-audioI2S/blob/5f3e30d0b3c61c48b81506545d665bbe4eb7936b/src/Audio.cpp#L2149
• https://github.com/dkm1978/ESP32-audioI2S/blob/5f3e30d0b3c61c48b81506545d665bbe4eb7936b/src/Audio.cpp#L2261

So, I rephrase my question: would it be possible to include the above changes in a proper way so that internal DACs also work smoothly?

[dkm1978]

Add #define INTDAC in audio. Cpp Dnia 01 lutego 2021 22:02 mettyw napisał(a): Hi, I am trying to use this lib with the internal DAC of an M5Stack Grey. At least for me, the changes in m_i2s_config aren't sufficient, I hear sound but it is extremely distorted. I could fix it by altering the function Audio::Gain() as well and add something like v[LEFTCHANNEL] = v[LEFTCHANNEL] + 0x8000; v[RIGHTCHANNEL] = v[RIGHTCHANNEL] + 0x8000; similar to what has been done here: github.com github.com do you think it is possible to extend your code accordingly? My understanding of C and bit-pushing is a bit rusty, so sorry for not providing a clearer fix directly. — You are receiving this because you authored the thread. Reply to this email directly, github.com view it on GitHub , or github.com unsubscribe .

[dkm1978]

See at 2261 line in my audio.cpp Dnia 01 lutego 2021 22:02 mettyw napisał(a): Hi, I am trying to use this lib with the internal DAC of an M5Stack Grey. At least for me, the changes in m_i2s_config aren't sufficient, I hear sound but it is extremely distorted. I could fix it by altering the function Audio::Gain() as well and add something like v[LEFTCHANNEL] = v[LEFTCHANNEL] + 0x8000; v[RIGHTCHANNEL] = v[RIGHTCHANNEL] + 0x8000; similar to what has been done here: github.com github.com do you think it is possible to extend your code accordingly? My understanding of C and bit-pushing is a bit rusty, so sorry for not providing a clearer fix directly. — You are receiving this because you authored the thread. Reply to this email directly, github.com view it on GitHub , or github.com unsubscribe .

[ranger2012]

I 've compared both codes from dkm1978 and from Wolly, it seems that dkm1978 source is much better, maybe somewhere else in the code there is difference...

[ranger2012]

setting the volume to maximum makes the sound clearer..

[mettyw]

setInternalDAC(true); //enables internal DAC

Hey thanks for adding this function and the flexible way how it is implemented is even cooler :)

However, there is still heavy noise using this lib compared to other libs. Actually, this can be avoided: I saw that you decided to add the increment before the call of the Gain() function. I think I understand the reasoning for that, considering the volume, filter & balance calculations that are done within Gain() . But according to espressif/arduino-esp32#2647 (comment) the purpose is to convert from signed to unsigned, since the internal DAC expects unsigned. So, I think the increment of 0x8000 should be at the very end before Gain() returns the result, regardless of the calculations that happened before.

If I do that change in my code, the output of the DAC is actually really clear (... well, for an internal DAC at least).

[schreibfaul1]

Hi mettyw,
that's it! Whether the output signal is distorted or not depends on the volume. The offset 0x8000 is now added after Gain (). Many Thanks!

[ranger2012]

sound is good now.. Thank you both for such a wonderful work..