fun couple of days in SF attending @Vapi_AI’s VapiCon! lots of builders and operators in the voice ai space all coming together to learn and share ideas. also seems like s2s models are gaining more and more traction
1
2
79
so good to be back in Van - we’ve got such a beautiful airport @yvrairport appreciate @united for getting us all here safely despite the craziness at sfo
103
nationwide ground stop for all flights in the US due to a comms issue - guess i might not be flying home today after all. wish me luck!
1
1
166
330+ notes on notesforbelal.site as of this morning!! I love that there are notes from all across the world from Japan to the UK to Brazil and North America 💙🤍
2
3
114
I made a site where you can write Belal Halbouni a note of encouragement, and every note fills a seat in a 3D BC Place 💙
The goal is to fill out all 54,000 seats so on hard days he can open it and scroll through a whole stadium worth of notes of strength and encouragement :) #VWFC @WhitecapsFC @MLS
notesforbelal.site
really love this isometric city trend going on at the moment. came across it for the first time a few months ago… wonder how much it costs to get a city online
1
1
226
been thinking a lot lately about cascaded voice systems vs native audio voice models that take audio in as well as output it and which will be more prevalent in the future…
cascaded models make sense when you look at them through the lens of where we started - they’re relatively easy to stitch together via different components like STT, LLM, and TTS and allow us to pick and choose the best component as each component improves. They’re also really beneficial for auditing text at every point in the chained pipeline which is great for compliance reasons.
However, my main concern with cascaded systems is the fact that they throw away at least half of what makes speech so important - all the paralinguistic context such as tone, prosody, emotion, etc… for me, personally, how something is said can sometimes be more important than what was actually said and I think we’ll want to build, and interact with, systems that capture both of those things in the future. If what we care about, as a society, is naturalness and AI systems that can steer their expressiveness to match your current state, then I can’t see us using voice models that lose the ‘how’ in the process.
Having said this, as with everything, it will also depend on the use case and context - task-oriented use cases will probably care more about what was said as opposed to the how and whether or not the task was completed whereas other use cases will most definitely want to understand how something was expressed in order to respond with the right tone and prosody.
1
1
177
oooo was there a claude desktop update? looks cleeeeean @alicelovescake1 :) i like it!
1
125
another awesome event last night put on by @DeepgramAI - had a blast talking voice and speech with everyone there. we’re so early!
6
289
awesome day full of learnings and meeting new folks - thx to @GoogleDeepMind, @kaggle, and @zapier for putting on a really well run inaugural event :)
4
20
670
the A220 is a lovely aircraft to fly on, especially for tall folks - the complimentary wifi onboard was awesome too @AirCanada!
84
will be my first time on an A220 this week - looking forward to seeing how it compares to the 737
72
I let one of my non-tech friends listen to this conversation and they failed to identify Dessa as the agent… pretty crazy stuff tbh
We made the first AI model you can have a realistic conversation with that was trained entirely on speech!
Dessa is a full-duplex speech-to-speech model that you can run on consumer GPUs.
This is a GPT3 moment for speech.
58



















