A few corrections

[Nielk1]

Here's some corrections based on my older research.

1. When the unk bool in the packet_t is true the length must be doubled (probably a pair of related data, maybe channels?).
2. There are special command packets that exist for example those that control the Mic.
   • Mic ON: 91 02 0b 1f
   • Mic OFF: 91 01 00
   • Mic ON (Media Remote): 91 01 01 (simpler than DS, though I think both command forms work on each)
   • Enable Keys (Media Remote): 91 01 02 (note this appears to be a bitmask, at least on the media remote, to toggle features. Thus 00 is all off, 01 is mic on, 02 is keys enabled, and 03 is mic on and keys enabled.)

I guess now that this is publicly known I should get off my butt and put the results of my research on my wiki (which you've linked actually, about 99% of it is my original research so I'm glad it's been helpful). I never got the audio out working because other life matters took priority, else we'd probably have had this information known sooner. I got hung up on the audio being Opus codec because that's what the microphone data is and had only considered other options like raw PCM just when I was no longer able to invest the time on the research.

Here are some packet samples that you can test against you packet structure. They are taken from a very terse BT capture which were all I could get due to limited tools and options. You can see the length doubling in practice here. packet_samples.txt

I'll try to find time to get the Opus data onto the wiki. It is true for both the DualSense and the Media Remote (Lotus). I have the media remote entirely figured out except for setting the IR codes, which is annoying.

[neptuwunium]

According to the psdevwiki, there's a way to select audio codec, but no clue if that means mic or audio.

probably both?

[egormanga]

Thank you for your feedback! I've somehow been missing this issue for a while.

I already had both microphone and speaker audio fully working at the time of publishing this POC, but didn't want to announce that for a bunch of reasons I covered in the Reddit thread.

As for the mic control packets, there are not entirely correct, and I will publish the details soon™.

Thank you for the samples, they will most certainly help us with the double packets.

[egormanga]

@yretenai That's probably referring to the hardware CODEC chip (i.e. audio controller), not the audio codec in software terms.

[NZX-DeSiGN]

Hi @egormanga,
do you have any updates on a potential release or publication of the microphone audio capture details?
I’m particularly interested in the mic side to experiment with building a custom voice assistant using the DualSense controller.

Thanks a lot for your work and for sharing your research with the community !

[egormanga]

I'm sorry, no particular updates on that topic. Real life, all stuff. I'll do as soon as I can; love making this project!

[egormanga]

@NZX-DeSiGN If you're onto something worth being called special, hit me up in DMs anywhere, I'll be sure to provide reasonable assistance.

[NZX-DeSiGN]

@egormanga thanks for the fast response!

I’m not sure whether my project can really be considered special — I’ll send you a DM and let you be the judge 🙂

I totally understand how real life comes first, and how fun and rewarding these projects can be, even if they take time !
I’ve personally had a lot of fun reverse-engineering my LG WebOS TV: capturing the remote’s microphone audio to build my own voice assistant and digging into the HDMI-CEC protocol to control my amplifier!

I'm really looking forward to the release of your final work, what you’ve done so far is genuinely impressive — great work, and thanks again for sharing with the community!