anyone have practical experience of USB over wifi

[fatoldgit]

SImple requirement.

Running Linux with alsa (pulseaudio/pipewire etc... all that crap is removed)

So if I run my playback engine on a laptop (files mounted over NFS from a build your own NAS.. files are all 44.1/16 flac) I assume if I use some USB over wifi device that I can configure the playback engine to use my DDC (i.e. the DDC USB port will be visible as it is now hard wired)

I havent looked much at what "stuff" is needed for USB over wifi but I assume there is a sender connected to my laptop and a receiver connected to my DDC?

Not worried about wifi speed... I will see ~1MB[ytes]... being 200k from NAS to laptop and 800k over the wif/usb.

Any one have any insight/recommended products?

Thanks,

Peter

 

[voodooless]

Isn’t this what streaming was invented for?

Why not have a pi and the DDC connected to it, and then stream to the pi. If you change to Pulseaudio or Jack, you can connect to these over the network as well. Alsa can’t do that.

 

[fatoldgit]

Isn’t this what streaming was invented for?

Why not have a pi and the DDC connected to it, and then stream to the pi. If you change to pulse audio or jack, you can connect to these over the network as well. Alsa can’t do that.

sure but if you have a "fat client" app without streaming then ....

 

[voodooless]

Or get a Chromecast Audio and use this:

GitHub - muammar/mkchromecast: Cast macOS and Linux Audio/Video to your Google Cast and Sonos Devices

Cast macOS and Linux Audio/Video to your Google Cast and Sonos Devices - muammar/mkchromecast

github.com

sure but if you have a "fat client" app without streaming then ....

Then you use Jack or Pulseaudio over network.

 

[fatoldgit]

Or get a Chromecast Audio and use this:

GitHub - muammar/mkchromecast: Cast macOS and Linux Audio/Video to your Google Cast and Sonos Devices

Cast macOS and Linux Audio/Video to your Google Cast and Sonos Devices - muammar/mkchromecast

github.com

Then you use Jack or Pulseaudio over network.

yep aware of this.

I like buggering around with stuff... fills in my retirement years so wanted to see what good usb over wifi is.

Thanks

Peter

 

[voodooless]

I like buggering around with stuff... fills in my retirement years so wanted to see what good usb over wifi is.

Well, essentially for audio, it's probably not. USB audio is rather timing dependant, and over a network, and especially wifi, all bets are off in that regard.

You could actually try, there is a way:

USB/IP - ArchWiki

wiki.archlinux.org

 

[fatoldgit]

Will have a crack with https://wiki.archlinux.org/title/USB/IP

there are a few other similar solutions but they cost money... not that I dont mind paying but Linux is FOSS after all plus if the above works I am self sufficient.

Peter

 

[fatoldgit]

Well, essentially for audio, it's probably not. USB audio is rather timing dependant, and over a network, and especially wifi, all bets are off in that regard.

You could actually try, there is a way:

USB/IP - ArchWiki

wiki.archlinux.org

didnt see your reply so great minds think alike!!!

one would imagine that while USB audio is time dependant, as long as the end point PC (the one connected into the DDC) has its buffers filled in time, then the timing is all local between the PC connected to the DDC and the DDC.

Thanks for your interest.

Will report back with results

Peter

 

[voodooless]

one would imagine that while USB audio is time dependant, as long as the end point PC (the one connected into the DDC) has its buffers filled in time, then the timing is all local between the PC connected to the DDC and the DDC.

For things like mass storage, yes that works fine. USB audio however is more delicate, even the less timing-dependent UAC2 asynchronous mode.

 

[tifune]

I used a Silex branded device for awhile, it worked fine with most of the DACs I tried as long as you didn't install 3rd party drivers. If your device is plug and play, and gives all the features you need (most DACs need drivers for 384k or DSD, for example) in that mode, it probably works OK.

 

[voodooless]

I used a Silex branded device for awhile

I doubt it has Linux support though.

 

[phofman]

Some time ago I played with USB-IP, it seemed to work OK over ethernet. Even the USB-IP client for windows worked (when made to work). Wifi is a very different beast, randomly hitting larger pings, much less reliable, much higher latencies and jitter.

This is a nice paper which shows the principles https://www.researchgate.net/public..._Extension_for_Device_Sharing_over_IP_Network .

As of async UAC2 - the host is expected to adjust the data flow based on the async feedback. I have not seen a spec how fast it should respond, there may be some. Maximum acceptable delay would depend on size of the internal buffer in the receiving USB device, probably very device-dependent. But a delay in the order of 10ms is probably OK, the rate-tweaking changes are usually very subtle.

For gigabit ethernet I believe it will be 100% OK. For wifi - the only way is to try and see performance of the actual setup.

The aforelinked paper talks about setting a queuing depth (how many URBs are transferred in one batch to/from the other network side). I do not know how to set this in the actual kernel driver/module, it would take analyzing a bit the usb-ip source code.

 

[Roland68]

yep aware of this.

I like buggering around with stuff... fills in my retirement years so wanted to see what good usb over wifi is.

Thanks

Peter

Also take a look at Dante, there you will find some information about WiFi transmission and problems.

 

[fatoldgit]

Some time ago I played with USB-IP, it seemed to work OK over ethernet. Even the USB-IP client for windows worked (when made to work). Wifi is a very different beast, randomly hitting larger pings, much less reliable, much higher latencies and jitter.

This is a nice paper which shows the principles https://www.researchgate.net/public..._Extension_for_Device_Sharing_over_IP_Network .

As of async UAC2 - the host is expected to adjust the data flow based on the async feedback. I have not seen a spec how fast it should respond, there may be some. Maximum acceptable delay would depend on size of the internal buffer in the receiving USB device, probably very device-dependent. But a delay in the order of 10ms is probably OK, the rate-tweaking changes are usually very subtle.

For gigabit ethernet I believe it will be 100% OK. For wifi - the only way is to try and see performance of the actual setup.

The aforelinked paper talks about setting a queuing depth (how many URBs are transferred in one batch to/from the other network side). I do not know how to set this in the actual kernel driver/module, it would take analyzing a bit the usb-ip source code.

Thanks for the info.

So my house was built in 2008 and being a IT guy it has structured wiring (including 48 CAT 6 ethernet ports in various locations with four RJ45's in my dedicated listening room... two directly behind my rack and two on the rear wall)

To minimize wifi issues, I will have a WAP running off the rear wall in my listening room so will have a strong signal.

Worst case, can run an ethernet cable to the laptop from the back wall.

Being a retired Unix/Linux systems level c/c++ programmer (along with other languages) I will be looking at the source code so if I have issues, I might be able to hack my way to goodness.



Peter

 

[fatoldgit]

I doubt it has Linux support though.

if stuff is UAC2 compliant then the OS "shouldnt" matter

will check out the device..thanks @tifune

Peter

 

[voodooless]

if stuff is UAC2 compliant then the OS "shouldnt" matter

will check out the device..thanks @tifune

The USB device is not the problem, the proxy device is. On the PC side, there is only software, and so far, it only shows Windows compatibility.