⚒️Thor, the Norseman⚒️ er en bruker på snabeltann.no. Du kan følge dem eller kommunisere med dem hvis du har en konto hvor som helst i fediverset. Hvis du ikke har en konto så kan du registrere deg her.

appear.in, a video conferencing tool that uses WebRTC (in-browser media compression and streaming), honestly has pretty bad sound and picture quality compared to Skype.

I tried using an appear.in room to stream my webcam back home on demand, but it sucked, so I made an extra Skype account instead and had my roommate log into it and enable auto-answer on it.

Risky? Nah. Nothing interesting happens in the living room when I'm not there. I just want to check on my cat from time to time.

@thor It's got noticeably worse on the free accounts since they started pushing premium.

As a premium subscriber, appear.in is great, but the best quality I've experienced is through Zoom (company I'm working for has accounts)

@dajbelshaw They lower the bitrate? Doesn't sound like a good sales strategy. When I talk about quality, and mention WebRTC explicitly, it's because I'm not only talking about the bitrate, but the quality of the noise filtering and echo cancellation, which is an issue with how WebRTC is implemented, not appear.in as such. The noise filter is too aggressive, and the echo cancellation is leaky. Skype seems to have much better algorithms.

@dajbelshaw When it comes to noise filtering in particular, I have yet to actually encounter a system that does a good job of it. Yes, the noise is filtered out, but so is half the speech information, and you end up with mushy audio and quiet sections of speech being filtered out. I always disable noise filtering if there's an option to do it. Skype doesn't seem to do any noticeable noise filtering, only very good echo cancellation, and that's the way I like it.

⚒️Thor, the Norseman⚒️

@dajbelshaw You sometimes wonder if the people who design these systems don't actually test them.

I have heard one good noise filter, and it's a very simple one that I designed myself. It doesn't do any frequency filtering. Instead, it checks if the input level is pulsing or not. Speech is rhythmic, so it passes through the filter. When there's no speech, it simply lowers the volume. The background noise is audible when there's speech, but at least you can hear the other person clearly.