Post 3194513 by kellerjustin@mastodon.social
(DIR) More posts by kellerjustin@mastodon.social
(DIR) Post #3194374 by jerry@infosec.exchange
2019-01-20T18:34:27Z
0 likes, 0 repeats
Idea: an always on alexa-style device that toots the text version of all conversations it hears in real time.
(DIR) Post #3194414 by m4iler@infosec.exchange
2019-01-20T18:35:37Z
0 likes, 0 repeats
@jerry Amazon beat you to that one, only they don't toot them.
(DIR) Post #3194464 by jerry@infosec.exchange
2019-01-20T18:37:18Z
0 likes, 0 repeats
@m4iler I think it should not only toot, but also provide narration so the toots are attributed to different people... i.e.,person1: where's my wallet?person 2: probably where you left it!
(DIR) Post #3194513 by kellerjustin@mastodon.social
2019-01-20T18:39:46Z
0 likes, 0 repeats
@jerryFound the cop 😆
(DIR) Post #3194663 by amiloradovsky@functional.cafe
2019-01-20T18:44:28Z
0 likes, 0 repeats
@jerryBillion followers. TV reality shows craving to reach that level of success.
(DIR) Post #3194713 by crazypedia@toot.chat
2019-01-20T18:46:42Z
0 likes, 0 repeats
@jerry @m4iler I mean, you might be able to do this with a pi zero 🤔
(DIR) Post #3194770 by m4iler@infosec.exchange
2019-01-20T18:49:30Z
0 likes, 0 repeats
@crazypedia @jerry Voice recognition on RasPi0? I don't know, sounds resource-intensive. But yeah, we need:1) web connection2) microphone (Banana Pro has one, AFAIK)4) voice recognition AND attribution
(DIR) Post #3194869 by crazypedia@toot.chat
2019-01-20T18:54:18Z
0 likes, 0 repeats
@m4iler @jerry the pi zero w can do visual recognition with on board tensor flow, G👀gle has a cardboard kit, it doesn't need internet to offload the processing, it uses a pre trained library. I didn't look but I'm sure there's libraries for voice too.You won't easily get named attribution without training, but you could have it attribute IDs to voices it hears like93a6f3: what's this836f3a: looks like a voice tooter93a6f3: so it's live tooting what we say836f3a: seems like it