Speech
topic on 3 shows · 3 statements across 3 episodes
Latent Space 
My First Million 
No Priors
3 statements about Speech, every show
Modern speech models operate autoregressively by adding waveform tokens to LLM vocabularies
“Speech is autoregressive. You effectively I mean, this was even back with, like, the Orpheus architecture a year and a half ago. You just add a bunch of waveforms to the vocabulary so that the LLM can output tokens that represent those waveforms, and then you …”
Speech AI demands functional accuracy while music AI requires emotional resonance
“Speech just needs to be right. Just like read me this New York Times article. And if it's a tiny bit non-expressive or a tiny bit robotic, that'll still get the job done. And the real creativity was happening in a totally different part of audio, which is musi…”
Grammarly research shows improved writing habits transfer to users' spoken communication
“When we do user research, we notice that people who use Grammarly repeatedly adopt patterns of communication, more effective patterns of communication, translated to non-written communication as well. So for example, if Grammarly keeps suggesting that you don'…”