2X Snapclients x 1 ESP32

[DerPicknicker]

Hey @CarlosDerSeher ,

this is just a discussion for my idea to create 2 instances of snapclients at one esp32.

My idea is:

An amp board has two outputs - each output should be independently used as snapclient because in Multiroom audio almost every speaker is a mono speaker.

This would reduce the cost and increases the flexibility. Also for the lyrat board. It has stereo output and then you can easily connect 2 speakers and have independent audio outputs...

This is just an idea which I came up with. Because I waste one channel per speaker one board with 2 independent speakers would be great. Then you can install one board and create different areas...

[CarlosDerSeher]

Things that come to my mind

Does the software need to mix LR stereo channels to mono or should it be able to chose L or R

An amp board has two outputs - each output should be independently used as snapclient because in Multiroom audio almost every speaker is a mono speaker.

Why not just make a stereo speaker then? What's the gain of having two mono speakers which, if I understand you right, will be located spacially close to each other and connected by wires to the DAC? Isn't this kind of contradictory to the sole purpose of having a WiFi speaker?

[DerPicknicker]

Yes your edit is correct. Two different streams.

[CarlosDerSeher]

This will complicate things even more. You'll also have to aggregate the audio data from two snapclients to a single i2s stream.

[DerPicknicker]

I think we can stop here ;-) getting more esp32+dac (or a raspberry with multiple dacs)should be much easier.

Would be a really Cool task for cold and bad days. @CarlosDerSeher are you now motivated 😅 ?

[CarlosDerSeher]

Maybe 😀 but this could take a while with my second child on her way. So don't wait for it 😉

[DerPicknicker]

No, No... I have firstly get my MA instance running on a new machine... ordering esp32 boards.

This is why I love the esp32 it doesn't have a file system .. So you can't break it. A power cycle and you're good to go.

Edit: ALL THE BEST FOR YOUR 2nd one 👍🏻

[CarlosDerSeher]

Also I can tell you this will be quite an adventure. You'll need to change all those global variables (decoder, audio buffer, etc) to dynamically created ones in the context of a snapclient module.

This will of course improve the code a lot and clean up most of the mess I created by "quick and dirty" making this thing do synced playback. Do you think you can do this? First things that comes to my mind is a seperate netconn connection not sure if we could just start http_task twice. I don't think so without refactoring big parts of the code.

[DerPicknicker]

I thought just double the player tasks and change a bit of the buffers. So I can do it but the problem is if you already say that it will be an Adventure then I think it will be more time consuming for me. It's not a skill issue - just a thing about priorities..

Because all I do with embedded is just my personal hobby. So maybe it's easier just get an raspberry and connect multiple soundcards to it.

If you see any other option to get this ready please let me know.

[CarlosDerSeher]

You'll also need a separate TCP connection (http_task) and a separate decoder (maybe others too don't knows right now). Not sure if the processor can handle all this, I think it could handle 2 though. Then there comes a decision, which of the time syncs you'll use. snapclient instance 1 or 2? Then you'll need the player task to aggregate both streams and make sure you pick the same point in time from the buffer. I don't think this will be something you'll come up with just like that. You could maybe just duplicate the player task and put all audio to a ring buffer with the size of one decoded chunk and do the i2s part from another task. Wait until both buffers have enough data and then start building i2s data.

[CarlosDerSeher]

So maybe it's easier just get an raspberry and connect multiple soundcards to it.

Why not get multiple ESP + DAC? Should be cheaper this way or am I wrong?

[DerPicknicker]

Lyrat costs around 20.. The esperagus HiFi costs 2x of the Lyrat and I guess it sounds much better due the lower noise.

Lyrat is horrible for audio. I can't find the time to finish the toslink pcb which will result in much lower noise.

A raspberry pi costs 20$ (for a zero). A usb hub (10$) and a 7ch usb soundcard 15$.

45$ for 7ch snapclients...