Hacker News Reader

Comment by kawfey | Hacker News Reader

This is cool. I can't wait to try it out!

Crazy Feature request: One of the things I've wanted for years out of a software based walkie-talkie like this is manual audio spacial diversity control (a.k.a. 3D audio). That way I can have multiple simultaneous channels, but be able to change the position (left/right/forward/45deg/etc) and the volume of each feed so I can have situational awareness with all channels while being able to use my brain to separate and focus on a particular channel.

These kinds of systems are often marketed to dispatch agencies, mission control, government/military customers, etc, and are super expensive[0]. I've never seen a consumer-grade version of 3D audio like this, but that would be super beneficial to ad-hoc communicators in disasters, scanner enthusiasts, public event coordinators, county/city-level EOCs that can't afford expensive systems.

[0]:https://www.thedrive.com/the-war-zone/24741/new-3d-audio-wil...

View on HN

Aha! I'm a huge fan of positional voice audio in video games. Mumble has a positional audio feature[0] that is relatively straightforward to integrate via mods. I built a mod[1] recently that adds support to Raft, and I attempted to list some of the benefits in the readme. In my experience, it completely changes the feeling of talking with people online; they become their avatars. I'd love to see it come to discord or zoom, and be lifted out of the video game space.

I haven't looked at Mumble's actual implementation, but I imagine it's possible with the web audio APIs.

[0]: https://wiki.mumble.info/wiki/Positional-Audio [1]: https://www.raftmodding.com/mods/mumble-link

I love that Mumble has that, but I wish it were possible/easy to enable it without a video game and manually position participants.

As for webapps, I think Mozilla Hubs does this: https://hubs.mozilla.com/#/

...as does Freeswitch I believe, though I've had a heck of a time trying to get that up and running: https://freeswitch.org/confluence/display/FREESWITCH/mod_con...

I would love for MozIlla hubs (or any similar app for that matter) to allow one to set a webcam feed as your avatar's "head".

If anyone knows of such a thing please let me know. Would be perfect for socialising during quarantine and beyond.

It's at the other end of the spectrum, but Remotion is really cool and does what you describe. We wanted to get away from always on video, but I know lots of people like that. I cannot say enough good things about what the Remotion team are doing.

Kosmo.io has a poker app where your webcam feed sits at your seat at the table.

You might be interested in https://www.calla.chat/ ( https://github.com/capnmidnight/Calla ) which supports spatialized audio based on participant location on a top-down map.

Edit: in some ways, this might be totally off-base since it does involve a game and participant self-arrangement. Hopefully it's somewhat relevant, though!

I could be wrong, but based on the README this looks like it just sets speaker volume based on position - which is neat, but is not quite positional audio.

I wondered about that too, and found https://github.com/capnmidnight/Calla/commit/abc851b49bd1801... before commenting which seemed to indicate that there's at least stereo positioning. It's possible I'm mistaken though.

Oh interesting, it looks like you're right! They do appear to be using a StereoPannerNode rather than the full 3D PannerNode[0], but that is more than just speaker attenuation.

[0] https://developer.mozilla.org/en-US/docs/Web/API/Web_Audio_A...

That's an old commit. It's had spatialized audio for the last month, at least.

I'm actually going to put a feature back in to let you select the audio "spatialization" model, because some folks have had trouble with PannerNode not working right. I used to have a selection between volume scaling, StereoPannerNode, and PannerNode. Been a little too busy at the day job to do it lately.

I have some multi-tenant PBXs based on Freeswitch however, I have not seen a positional audio feature in Freeswitch. Can you give more insight? Your link is the conference module and unrelated I believe.

I made a proof of concept chrome extension a bit ago: https://github.com/anateus/amphisonic

Currently re-writing it to be actually decent code. Basically, it takes up streams on the page and arranges speakers in a semi-circle in front of you, letting you adjust the radius (I've found for particular numbers of speakers there's an optimal radius for speaker separation).

The now defunct Dolby Axon had this feature for up to 50 people. It was great, you could drag people in a grid around you (see picture in [1]). I’m still sad they discontinued it, it’s one of the few voice chat services that I would gladly pay money for.

[1]: https://venturebeat.com/2009/09/04/dolby-launches-axon-surro...

Aureal A3D. Technology was developed in the nineties for NASA. Stolen by Creative. Creative simultaneously stole their patented tech and sued Aureal for bogus patent infringement. Aureal "won" the Lawsuit, but ran out of money due to legal fees. Creative bought their assets, IP and funny declaration releasing them from any damages from lost patent lawsuit.

Patents long expired, nowadays smallest $2 microcontroller can run it in real time.

Check out https://www.highfidelity.com/

I know it's not really what you are asking for, but it's some very cool spatial audio work, letting you separate out where folks are in a virtual space.

[disclaimer, I know some of the folks working on it]

I tried out highfidelity a couple weeks ago and didn't get any of the directional audio to work. There was a radial but not an angular effect if that makes sense.

interesting. It was working for me on Chrome, I could differentiate left/right

Virtual Airwaves is working in this space.

https://virtualairwaves.com/

and see their iOS app: https://appsto.re/us/jrj3hb.i and the web app: https://cb.virtualairwaves.com/channel/1

They have several patents on "3D Audio" style techniques to give distance cues via the signal:

See:

https://patents.google.com/patent/US9730023

I wish podcast producers would do this (or more of them at least in case anyone knows of one). I’ve done it manually as an experiment and it definitely improves clarity in cases where there are 4 or more drunk folks taking over each other.

Thanks! The tech in the link is awesome...might be above our pay grade.

It's definitely doable. Dolby Axon (which shut down a few years ago) had positional voice chat.

It should be easy to integrate game audio engines, such as OpenALSoft.

Reply on news.ycombinator.com