Today, we’re excited to introduce Miso One, the most emotive voice model in the world.
Miso One is an 8-billion-parameter text-to-speech model for highly expressive speech generation. It emotes like a human and responds faster than a human, with just 110 milliseconds of latency.
We’ve open-sourced the model weights, with API access coming soon.
Hear how Miso One sounds in the thread below.
placing my bets for some evil HR company to make an AI agent that announces redundancies to employees for you, extra points for AI “avatar” who gives you an exit interview
@VahidK What a weird angle to spin this? The government are the very people putting your soldiers in wars.
Giving the government the ability of autonomous weapons or mass surveillance puts a lot more capability in the hands of people who have historically not given a shit about you.
agi will be achieved the second that claude checks for the presence of package manager lockfiles to choose which tool to use
i refuse the notion that i should have to tell it specifically in a claude md file
@Clad3815 Any reason why you opted to include reading game state from RAM? That’s often privileged data that’s not generally accessible & not transferrable to other games or even other instances of the game in a lot of cases.
cursor, claude code & claude desktop all being horrendously buggy, whilst being some of the most prominent vibecoded software is unsurprising
maybe 27 prs a day isn't sustainable or optimal
Would be interesting to see if y’all could inject contextual HUD elements, a lot of games give a small text prompts for “F to pay respects” or similar, so I imagine the training data exists for it.
Could probably also help in extremely fine-grained action trajectories if there’s a text representations which interact with some learnt “interact” discrete action. Thinking along the lines of dialogue/action options in things like Baldurs Gate.
@shlomifruchter did y'all end up using a lot of the diffusion modelling from GameNGen in Genie? 1 had no diffusion but 2 & 3 had huge uplifts in visual fidelity
@bcherny Probably two separate issues, but yeah the extreme slowdown/halting has been annoying, I’ll try and get a heap snapshot & profile & I’ll /feedback it
Meet Sprout.
Today, we’re releasing a new kind of robotics platform. One designed to move out of the lab and into the real world, closer to the people who will shape what robots become next. [1/4]
this feature it’s something that blows my mind that it doesn’t exist in @Jira or @SlackHQ
we have a primarily Australian team but with Canadian/US team members, makes communicating times such a manual process for everyone