We've launched Callout: make AI calls in English or Japanese.
Product Hunt:
producthunt.com/products/callo…
English
Maxim Makatchev
437 posts
Maxim Makatchev
@maxipesfix
founder of https://t.co/yK3uD96q4s AI's next UI. conversational AI blog: https://t.co/fACup81SvP
San FranOsaka Katılım Temmuz 2014
1.3K Takip Edilen368 Takipçiler
@pipecat_ai is our inspiration as well as the upstream Smart Turn repo so, the shout out from @kwindla is much appreciated!
kwindla@kwindla
.@maxipesfix forked the open source audio Smart Turn model and added video! Smart Turn is a "turn detection" model, used in a conversational agent to decide when the agent should respond. The model, training data, and training code are all completely open source. When we built the first version of Smart Turn, enabling this kind of extention and collaboration is exactly why we wanted to make everything open source. Maxim's blog post is super useful to read if you're interested in training multimodal models. It describes the design choices and technical details (3D ResNet, late fusion, two-stage training, inference runs on GPU in ~100ms). And all the code is available in the GitHub repo. Really great work.
English
@ashlleymo @kwindla Thanks! May I vote for a conversational video dataset? :)
English
.@maxipesfix forked the open source audio Smart Turn model and added video!
Smart Turn is a "turn detection" model, used in a conversational agent to decide when the agent should respond. The model, training data, and training code are all completely open source. When we built the first version of Smart Turn, enabling this kind of extention and collaboration is exactly why we wanted to make everything open source.
Maxim's blog post is super useful to read if you're interested in training multimodal models. It describes the design choices and technical details (3D ResNet, late fusion, two-stage training, inference runs on GPU in ~100ms). And all the code is available in the GitHub repo. Really great work.
English
Maxim Makatchev retweetledi
I built a tool that researches ideas so you don't waste time building bad ones. Launching early access today (docs coming soon). Quick research + real citations. Free credits for all, bonus for early birds. Sign up, DM/email me your email. Link in the next tweet.⬇️
English
Last week, just a day before Gemini 3 was released, @susuROBO helped run a 1-hour version of the @pipecat_ai x @GeminiApp hackathon at Osaka College of High Technology @osaka_hightech 大阪ハイテクノロジー専門学校.
We used Pipecat's SmallWebRTC Prebuilt repo as a starting point (thanks @kwindla and @aconchillo!), which allowed even freshmen to finish the hour running a multimodal voice agent on their laptops.
Ironically but not surprisingly, what they've built in under an hour was in a few ways more advanced than the Alexa they got as prizes.
日本語
Dawned realization:
You can only sell a product to someone whose likes you can get to know in person.
If the only people you meet are SEs and PMs, you’ll likely only be able to sell your developer time, or perhaps a productivity tool.
English
@amir_harati It might be a small task, but how do you know it's the correct change?
English
@maxipesfix It was relatively small task :) and get complaints but you are right on priorities here.
English
Some UI/UX (including font size ) are updated. I will add some help and update the landing page next.
Maxim Makatchev@maxipesfix
@amir_harati Unfortunately, the choice of having your largest fonts nearly as small as X's smallest fonts leaves out me as a potential user.
English
Talk to your customers lol @maxipesfix
Ryan Grim@ryangrim
Imagine you’re Chuck Schumer watching this
English
@RockZhang Let me put it out there that Notta is the single most used and useful service for a foreign entrepreneur in Japan.
English
@amir_harati Unfortunately, the choice of having your largest fonts nearly as small as X's smallest fonts leaves out me as a potential user.
English
Well that was a fun day and night of programming: Integrated Gemini Live Multimodal for voice convo, Gemini Flash for gesture recognition, and AIROID avatar into a @pipecat_ai pipeline for non-verbal user gesture recognition/response:
youtu.be/0yOsHS209ww
@AITinkerers
YouTube
English
Jump-starting an ecosystem takes a confluence from many directions. We timed the first AI Salon Kansai so that it coincides with @EDCON_Official and Global Startup Expo last Thursday. The result was a mix of local and international builders that reminded me what was great at meetups I joined in SF.
English
Come to think of it, receiving some correspondence in Esperanto would be a nice touch.
English
It's almost a rule that if a live demo does not fail in some way it's either not live enough or not revolutionary enough. So they are in a good company of Gates demoing plug-and-play and Jobs demoing a USB camera or iPhone youtube.com/watch?v=znxQOP…
YouTube
near@nearcyan
feeling really bad for the Meta OS team
English
Numazu city’s official VTuber sparks backlash after criticizing businesses for using generative AI imagery.
ITmedia AI＋@itm_aiplus
画像生成AI利用の店とは「距離を置く」――沼津市公認VTuberの“AI反対宣言”が物議 「見解異なる」と同市 itmedia.co.jp/aiplus/article…
English