ElevenLabs launched the v4 and v4 Turbo speech models on Monday, featuring lower latency, extended inline expression tags, and support for over 90 languages up from 70. The updated architecture allows users to clone any voice using a 10-second audio clip while improving voice identity consistency ac