Chat : 25 April 2022

16:09:05 From Mark Anderson : Yes, I’ve been enjoying seeing the stuff posted in the Twitter group.
16:11:14 From Fabien Benetou : with a location known already?
16:12:08 From Fabien Benetou : on the metaverse and federating Hubs I recorded
16:13:33 From Fabien Benetou : Peter for voice running on device or your own server I had moderate success with at least quite convenient to setup with web clients.
16:13:55 From Peter Wasilko : Thanks Fabien!!!!
16:14:45 From Fabien Benetou : quite ironic that we might need a white room to enjoy VR privately…
16:16:10 From Brandel Zachernuk : Nuance
16:17:26 From Mark Anderson : Nuance owns Dragon and bough MacDictate. All goes back to a single code base that has bankrupted several owners across the years.
16:18:17 From Mark Anderson : Previous owners include Lernout & Hauspie (sp?)
16:23:04 From Peter Wasilko : I remember using MacDictate
16:24:40 From Peter Wasilko : One thing I always found irritating at about speech systems was there wanting to work with a simple fixed command grammar but not letting you add new utterances that can be matched on the fly.
16:25:57 From Peter Wasilko : I’d want to be able to say “Recognize ” and have the system treat that audio fragment as a unique identifier to match on sound without trying to do any further processing on it.
16:26:33 From Peter Wasilko : Then use that “name” as a key to access structured data.
16:26:36 From Fabien Benetou : my explicited worldview motivated by computational comparison
16:26:48 From Fabien Benetou : (inspired by research from Clement Vidal)
16:26:56 From Frode Hegland :
16:29:25 From Mark Anderson : @ Peter I has an erstwhile PhD colleague doing the worlds first spoken Arabic dataset. What I learbed was the process isn’t at all what you intuit. It works on very tight snippets of sound and not words or phrases as we might think. Still all just probabilities at the bottom.
16:30:20 From Fabien Benetou : glTF
16:31:53 From Peter Wasilko : I found project that basically had it working about 8 years back and I wrote to the prof about the idea of an audio-memex, but he only wanted to sell his tech to a major Record Label; he was totally locked into the idea of the Big Music as his only potential cutomer/exit, so he’d get one Mega Check to sell it off off so he could retire to a tropic island somewhere!
16:38:29 From Fabien Benetou :
16:38:30 From Mark Anderson : Not, I think
16:39:58 From Fabien Benetou : his recent book on VR
16:44:36 From Mark Anderson : A timely XKCD:
16:46:20 From Peter Wasilko :
16:48:22 From Peter Wasilko :
16:55:40 From Fabien Benetou : sure it’s a monitor in the same way you can ride a Formula 1 car to go buy groceries…
16:58:27 From Peter Wasilko : I’d really like your doing Blender!
17:13:30 From Mark Anderson : I seem to be about 18” ‘wide’
17:29:51 From Fabien Benetou : I have a XS, I feel offended.
17:30:04 From Frode Hegland : 7 was the most beautiful one. By far
17:36:52 From Mark Anderson : TIL I have phone no one wants 🥲
17:38:57 From Fabien Benetou : I posted an image this morning on going from Blender to Spoke (and then to Hubs)
17:40:55 From Frode Hegland : thanks
17:42:36 From Brandel Zachernuk :
17:43:13 From Fabien Benetou :
17:44:10 From Fabien Benetou : (I mention this because publishing this way to a platform that we can customize means insuring than an extended format e.g glTF can include extra meta data to be parsed)
17:46:12 From Fabien Benetou : (e.g )
17:51:11 From Fabien Benetou : (another example parsing the export to modify while running )
18:03:49 From Peter Wasilko : Me too,

Leave a comment

Your email address will not be published.