sofiechan home

Sofie Calls, "spaces" on Sofiechan - feature request

anon 0x3b9 said in #2288 3mo ago: 88

Could be interesting to have some form of "twitter space" / broadcast audio feed for these conversations.

I was thinking some way to schedule calls based on time slots and topics, both callers have voice-distortion, call get's recorded and put on the feed - audios get ranked similarly to text posts and low-performing recordings get deleted.

Just an idea.

Could be interesting 88

anon 0x3ba said in #2289 3mo ago: 22

I love social audio, but my sense is that it's technically hard to do well, much harder than anything with text. And using voice distortion etc. would make it harder.

But if you only post the recording anyway, maybe there's an easier way. Have a chat stream at a scheduled time, capture the text transcript, then run it through an AI voice synthesis service to produce the audio that gets posted. No need for voice distortion, since the synthesized voices will be different anyway.

I love social audio, 22

anon 0x3bb said in #2290 3mo ago: 22

For some reason, I find Twitter Spaces disconcerting. It is the same for Zoom calls. I wish there was an alternative. I prefer the completely disembodied nature of conference calls, which I never used as a professional, but only as a kid, calling into hacked numbers. They more easily allowed anonymity, distortion, and interruption.

For some reason, I f 22

anon 0x3bd said in #2293 3mo ago: 55

I have also come to prefer voice over video. But likewise i prefer text for many things. Careful thought in dialogue is best done in public mid-length text exchanges ime. Thats why places like this are good. But maybe there is new kinds of fun and profit to be had with anonymized voice. With all the new AI tech it would be funny to transpose voices and cadences into the characteristic style of famous figures. Thiel vs mr Rodgers vs Putin on the subject of best girl in the latest anime. True voice anonymity requires a lot more style transposition than text, not just voice modulation. Usually you can tell who is speaking even if it’s shifted around. Full AI style transfer might fix this.

That said i have no immediate plans for voice. —admin

I have also come to 55

anon 0x3f7 said in #2400 2mo ago: 33

what about a discord server? Or is that too... identifying?

referenced by: >>2403

what about a discord 33

anon 0x3bd said in #2403 2mo ago: 44

>>2400
Lets just hang out in person, anon. Discord feels like they want to turn you in like the stasi. I wonder how hard it would be to get claude to build a basic webrtc syatem for sofiechat though.

referenced by: >>2404 >>2405

Lets just hang out i 44

anon 0x3bd said in #2404 2mo ago: 33

>>2403
Thinking a bit more about this, the thing to use would be Jitsi, which is an open source webrtc app like zoom. We could just set up an official sofiechan jitsi and then integrate it into the sofie UI progressively and experimentally. Anyone have good arguments for why we should or should not do this?

Thinking a bit more 33

anon 0x3f7 said in #2405 2mo ago: 00

>>2403

matrix? dunno if it has voice chat, but it is decentralized supposedly

matrix? dunno if it 00

anon 0x406 said in #2434 2mo ago: 11

Distortion would be a really cool technological problem to solve - and it's quite easy actually with the folks I know.

Distortion would be 11

You must login to post.