Outside the box: Using Murmur/Mumble as in-ear monitoring

[Oceanwatcher]

I hope this is the right board to post it on as this is definitely outside the box!

During research for making an IP based in-ear monitoring system, I stumbled across Murmur/Mumble.

In-ear monitoring is basically monitoring for musicians using ear plugs instead of loudspeakers (wedges) on the floor.

This has many advantages - no feedback, each band member can (if the system allows it) get a separate mix, controlling audio levels means protecting your ears by blocking out loud stage levels etc.

But it requires low latency - preferrably lower than 10 ms. Actually lower than 5 ms would be good.

So I am asking the collective brain out there:

Is it possible/practically doable to set up a minimal linux pc with Murmur and Mumble running and use Aandroid/iOS clients over Wifi to listen to a single audio source?

This would be a one-to-many over Wifi system most of the time with the occasional message back or message from backstage to bandmembers.

Is there a way to feed audio from an ASIO audio card straight into Murmur without using a Mumble client?

Is Murmur/Mumble still being developed? I noticed that the Android client was very old, or did I misunderstand?

Is this the best candidate for my project, or do you know of a better solution?

I know there are a lot of indie bands that would be thrilled to have a really low cost in-ear system available. Most musicians these days do not earn a lot. But they still deserve a decent monitoring system. So if this could be solved by using a cheap pc, a wifi router and their smart phones, it would mean a lot!

[fwaggle]

I'm pretty sure even without Mumble in the mix, you're not going to get sub-10ms latency from microphone to earpiece using a PC and smart devices. Wifi alone is going to add notable jitter to the signal, not to mention the unavoidable delays from the operating system - which while they're getting better are not non-zero, and you're going to have two of them.

The best approach, IMHO, would be to take something like Opus and see if it's applicable to building into an FPGA (my DSP-fu is seriously weak, I've no idea whether this is possible, done before, or what). Feed that into a naive digital radio transmitter that doesn't do any retransmission or anything, then catch it and decode it at the other end. If you lose a frame or get bit errors or whatever, Opus will handle it at least somewhat gracefully (to a point), and will let you transmit at a lower bitrate.

There's probably a reason why good wireless monitor gear isn't cheap.

I don't think Mumble is the solution you're looking for, but some of the tech it's built on might be. I do know of a few people who use Mumble to jam remotely, and in that way it works (for some definition of "works") but for such close proximity as monitoring live audio at a gig, there's just going to be way too much unavoidable latency.