26 ms·
I was wondering about that. * Regarding the VPN: Inevitably it adds an extra hop but with a VPN that provides sufficient bandwidth, low latency, and sufficient
by hackuser 11y ago
I was wondering about that.
* Regarding the VPN: Inevitably it adds an extra hop but with a VPN that provides sufficient bandwidth, low latency, and sufficient processing resources to decrypt/encrypt at wire speed -- I assume that either it will cost extra or will be something the user has to setup at a good hosting provider -- couldn't performance be sufficient for voice? IIRC, from long ago, voice needs ~80 Kbps.
* Regarding cellular data: Cellular connections are very widely used for voice, of course. Cellular data connections are used now (e.g., VoLTE). On one hand I'd have the same doubts you do; on the other it seems to work. Aren't there already VOIP apps?
- e12e 11y ago"Uncompressed" voice "needs" 64kbps (hence ISDN was 2 voice channels etc). If you use a modern codec, you could probably get away with less, for high-quality mono voice. Note that you might not want to use VBR, for other reasons (I thought this was already submitted to hn - couldn't find it - so here's a fresh submission): "Attacks on packet length may be surprisingly good: Hookt on fon-iks": https://news.ycombinator.com/item?id=10462363 https://news.ycombinator.com/item?id=10462363 That said, I suspect that eg: Ogg Vorbis at 64kbs CBR would be fine for voice, assuming you could keep the latency down. Latency and drops would probably be the biggest issue. And you can of course prove, that latency sets a limit to the number of hops -- say if you'd want to hop through a European jurisdiction from the West Coast. I'm actually planning to (try to) move to a pure data-oriented (not sure if there will remain a cellular part, or if I'll just use wlan) (probably) SIP for all my phone needs. That'll probably involve a SIP server on dedicated (rented) hw in Germany, and VPN from the phone/laptop. Not quite sure yet if it'll work out, or if it'll be good enough.