Block's Dessa is a voice model trained only on recorded speech, with no text at all
Block says Dessa is full-duplex, learned from human conversation alone and runs on an M4 Max MacBook Pro. Safety researchers get early access first.
Topic · 5 stories
5 stories tagged voice-ai on AI Tamer, newest first.
Block says Dessa is full-duplex, learned from human conversation alone and runs on an M4 Max MacBook Pro. Safety researchers get early access first.
Twilio outbound media expects base64-encoded raw μ-law audio at 8 kHz. A voice model’s PCM or WAV output needs explicit conversion before it reaches the call.
One missing IP fragment can erase a whole UDP datagram. Path MTU discovery and application-sized messages help voice systems avoid that failure mode.
Read RTP sequence numbers and timestamps together to separate likely packet loss from suppressed silence, then account for comfort-noise packets.
WebRTC is built for live media; WebSockets stream audio as ordered messages. Here is how the transport choice affects latency, loss, browser audio, and deployment.