webjack
webjack copied to clipboard
publiclab
loopback in browser
(very cool library, looking forward to using it!)
i think the loopback is already there, it's processing audio from the microphone, it just doesn't decode it correctly yet :) I tried the master and fix-decoder branches. Does it work for you?
I haven't gotten it to loopback in the same browser or between mic and speakers. Is there any filtering in the audio processing for suppressing echo?
I'm interested in if it's possible to pass data between webpages this way.
On Aug 5, 2016 4:32 PM, "Mark Vayngrib" [email protected] wrote:
(very cool library, looking forward to using it!)
i think the loopback is already there, it's processing audio from the microphone, it just doesn't decode it correctly yet :) I tried the master and fix-decoder branches. Does it work for you?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/publiclab/webjack/issues/24, or mute the thread https://github.com/notifications/unsubscribe-auth/AABfJzOkGlsGT6pr20G28eWecBl4gLh6ks5qc53igaJpZM4JeCU8 .
I don't have my notebook with me, so I can't set up the gh-pages of my fork with the fix-decoder branch. Browser to browser transmission should do better with that fixes.
Seems like the web audio API has echo cancellation build in. Eventually it can be turned off: http://stackoverflow.com/questions/31306309/how-to-disable-web-audio-analyzer-filtering-high-frequencies
Interestingly I could observe loopback even with a cable attached, when using with SoftModem. There is crosstalk between the signal lines. Tried to get rid of it by sending the inverted signal on the other audio channel, but without success. I can post a picture of the crosstalk tomorrow. Any ideas on how to suppress this loopback are very welcome :-)
Just tested loopback with echo cancellation turned off, but it did not make a big difference.
When I placed the loudspeaker approx. 5cm away from the microphone and set the volume really loud, it did loopback to the same and to other tabs.
However, with a cable connected (just one end, the other one is unconnected) the loopback is almost perfect :worried:
I'm guessing the latest gh-pages does not have the fix-decoder branch merged, so this may not be any new info, but I tried this out again with browser-to-browser and got some transmissions (holding speaker to mic) but pretty inconsistently (under 20% maybe).
OK -- go ahead, and I can use that in my tests (back in town now and have access to things/tools/soldering). Thanks!
On Mon, Aug 15, 2016 at 5:40 PM, Richard Meister [email protected] wrote:
Correct, fix-decoder is not merged yet, but we can do that now.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/publiclab/webjack/issues/24#issuecomment-239854723, or mute the thread https://github.com/notifications/unsubscribe-auth/AABfJy62nAfYBxJrDLGvVdtI8zkKxB86ks5qgJZjgaJpZM4JeCU8 .
Posted same links to another issue, but in case it's interesting / useful -- here's an example of browser looping back into browser with another library / codebase:
https://cho45.stfuawsc.com/WebAudio-Modem/FSK/modem.html https://github.com/cho45/WebAudio-Modem/tree/master/FSK
