Open Source Platform Projects at ASR

[Rick Sykora]

This is a place holder thread for high-level discussion about the vision for platform of open source projects.

The genesis of this platform came from a late 2020 discussion over the identified need for a high quality DIY active speaker and a need for a streamer to support content for it.

The current projects are:

1. Directiva Speaker Project
2. (TBN) Streamer Project

This thread is to propose new projects and cross project discussion.

 

[somebodyelse]

Highly synchronized wired/wireless network endpoints came up in passing in both threads, but are probably almost a separate project. Whether it's something using a standard protocol like AVB (not wireless) or AES67, or something less standard like gstreamer's time sync, I guess would be to be decided. I think doing it wired is relatively simple - that's not to say it's necessarily easy to make our system, but PTP works on wired networks and there are already standard protocols that are known to work. Wireless is another matter - PTP is known to have issues over WiFi, but there may be something in WiFi6 that changes that. gstreamer's old sync method was shown to be more reliable than PTP over (old) WiFi, but I don';t know whether it's good enough for stereo from different endpoints.

 

[somebodyelse]

A physical remote control with buttons and rotary encoder(s), and maybe an oled display, for those who don't like touchscreens or phone apps. I've proposed something similar before for using the MOTU AVB series interfaces as a preamp, using their network API, but the same hardware could equally be used with LMS, MPD and probably others. There are boards combining an ESP processor(hence WiFi and sometimes Bluetooth), battery charging, and sometimes an oled display (and possibly accels?) for ~$10-15 on ali, ebay etc. The project would supply a reference hardware control board for buttons and encoders, and an easy way to map them to remote control commands on whatever we use for the streamer project plus possibly others (kodi? Roon?). Similarly we'd need info for display if one is present. Bonus points for a charging dock with pogo pins instead of the usb connector, or wireless charging.

 

[StefaanE]

Highly synchronized wired/wireless network endpoints came up in passing in both threads, but are probably almost a separate project. Whether it's something using a standard protocol like AVB (not wireless) or AES67, or something less standard like gstreamer's time sync, I guess would be to be decided. I think doing it wired is relatively simple - that's not to say it's necessarily easy to make our system, but PTP works on wired networks and there are already standard protocols that are known to work. Wireless is another matter - PTP is known to have issues over WiFi, but there may be something in WiFi6 that changes that. gstreamer's old sync method was shown to be more reliable than PTP over (old) WiFi, but I don';t know whether it's good enough for stereo from different endpoints.

There is of course WISA, used by folks like Buchardt Audio.

 

[somebodyelse]

There is of course WISA, used by folks like Buchardt Audio.

There is, but last I looked it wasn't sufficiently open for us to interoperate with in DIY projects. This may be by design, in the name of copyright protection, so that AVR and surround speaker manufacturers would be able to adopt it without violating their HDCP license.

 

[m_g_s_g]

I was hoping some more progress in wifi6 to support PTP. From this paper, it seems that much of the effort to improve low latency has been postponed until wifi7 ...

A couple of years ago (I was still into drones then, before crashing a couple of them) I stumbled into this very interesting project/idea. Instead of using the full Wifi protocol stack to stream video data, this guy puts two (or more) Wifi dongles in monitor / promiscuous mode and the sender just broadcasts the video (this was intended to transmit video from a drone to a FPV first-point-of-view headset). I built a complete system around that with a Rpi A+ onboard.

I can’t avoid thinking that many of the issues associated with video transmission are shared in audio and that maybe this idea could be useful In that context. I wonder if that’s how Sonos sends the subwoofer and surround channel data.

 

[somebodyelse]

Wifibroadcast is certainly interesting. It's got some properties in common with WiFi Multicast which has its own set of problems that won't be addressed any time soon. I'll have to check on wifi device compatibility. I have wondered whether Sonos use something at a lower level than the network for sync, or whether it's some out of band signal they're using like another license exempt radio pulse.

 

[Rick Sykora]

Hi all,

If you have read the streamer project thread, you now know we lost the project lead. In the short term, I will try to fill in, but Directiva is at a critical stage too and so can use some help. The role is voluntary and mainly a technical liaison one between the contributors and I.

If you are interested and can help, please let me know.

Thanks,

Rick

DIY Leader
Audio Science Review

 

[Karu]

Many people want an open source equivalent to the W371…

 

[bigjacko]

Hi all,

If you have read the streamer project thread, you now know we lost the project lead. In the short term, I will try to fill in, but Directiva is at a critical stage too and so can use some help. The role is voluntary and mainly a technical liaison one between the contributors and I.

If you are interested and can help, please let me know.

Thanks,

Rick

DIY Leader
Audio Science Review

What happened to previous project lead? Did he quit?

 

[somebodyelse]

What happened to previous project lead? Did he quit?

https://www.audiosciencereview.com/...pen-source-streamer-project.20840/post-809110

 

[pozz]

Not exactly in the spirit of this thread, but thought these third party projects @RickSanchez posted were worth mentioning:

• https://www.audiosciencereview.com/...-raspberry-pi-w-raspbian-64-bit-config.15113/
• https://www.audiosciencereview.com/...m-support-for-android-and-raspberry-pi.14974/

 

[syn08]

I can only dream of an open source UAC2 stack, on a STM32 with USB hardware, if possible. There are simply no viable options today for an UAC2 stack, XMOS is an abominable architecture, a power hog, the source code is not available, and guarded by licensing, CMedia is not delivering even the data sheets or an SDK (for the 6635), the TinyUSB project still doesn’t have a working UAC2, the Borgestrand audio widget doesn’t look to me as portable (I might be wrong, though, didn’t look very closely to the sources), etc... long story short, there’s no open source safe place to go.

I’m trying to initiate such a project for quite some time, unfortunately I noticed very little traction for such a highly specialized and niche project, there are other more exciting areas that get the community resources. If somebody could assume the project management, I would definitely volunteer myself in. After a 40 years career in software (from code warrior to upper management) out of which 20 were in embedded software, I think I could bring something to the table. Completing such a project under the ASR moniker would be IMO a great legacy.

 

[somebodyelse]

Why try to initiate a project rather than contribute to an existing one like TinyUSB? Is there some issue with their code, licensing or development practices?

 

[syn08]

That's something to consider, although IMO is much easier to start from scratch with a dedicated project, then merging in a project that tries to cover everything and then something on top. And, after skimming through the tinyUSB code and docs, it doesn't look like having much room for UAC2 specific optimizations, plus that it doesn't seem to be targeted towards supporting the full UAC2 spec (appears they sacrificed some performance vs. extending the range of MCUs and boards). This is not a technical critique, the work done is wonderful, but more like a project scope issue.

The sad part, in all truth, it is probably too late to focus on either tinyUSB or any other UAC2 stack. In 1-2 years, probably before such a project gets the first UAC2 stable results, UAC3 (with its way to handle data polling, no more "pinging" the USB 2.0 interface) and/or Thunderbolt (daisy chaining, low latency, upsampling if needed) will already be everywhere. I would love to know in advance the winner .

 

[hochopeper]

The audio-widget project went pretty close but the lack of multichannel support is regrettable (really a limitation of the mcus available at the time the project started). Something similar based around stm32 would be amazing!

I think uac3 seems more likely given that a thunderbolt device would be effectively a pcie device and mcu manufacturers seem more likely to move USB devices.

 

[antcollinet]

Open source subwoofer? Possibly with 2 or 3 different sizes/cost levels.

 

[Madjalapeno]

Open source center channel? Possibly with 2 or 3 different sizes/cost levels.

 

[dearchap]

I can only dream of an open source UAC2 stack, on a STM32 with USB hardware, if possible. There are simply no viable options today for an UAC2 stack, XMOS is an abominable architecture, a power hog, the source code is not available, and guarded by licensing, CMedia is not delivering even the data sheets or an SDK (for the 6635), the TinyUSB project still doesn’t have a working UAC2, the Borgestrand audio widget doesn’t look to me as portable (I might be wrong, though, didn’t look very closely to the sources), etc... long story short, there’s no open source safe place to go.

I’m trying to initiate such a project for quite some time, unfortunately I noticed very little traction for such a highly specialized and niche project, there are other more exciting areas that get the community resources. If somebody could assume the project management, I would definitely volunteer myself in. After a 40 years career in software (from code warrior to upper management) out of which 20 were in embedded software, I think I could bring something to the table. Completing such a project under the ASR moniker would be IMO a great legacy.

@syn08 I'm with you on this. I have tons of embedded software experience and would like to do something like this

 

[pierre]

I have in mind to do an open source upmixer. If you are interested please like the post so I can guess interest.
It is relatively easy to provide as a VST plugin. I wrote a 2.0 to 5.1 one some time ago and it would be relatively easy to go 7.1.6
or whatever you fancy but not atmos since it requires proprietary encoder/decoder. I have plugins to do the encoding but that's ok for testing not for distribution. There are many ways to do a upmixer. From my experience using Penteo, you can do a good job per
track but it is hard to find a set of parameters that works well for all the tracks in an album. Possible some AI could help there.

Feedback?