When are open-source voice models coming?
AI voice solutions right now:
open-source models (Whisper)
user-provided fine-tunes - FakeYou, ElevenLabs (often bad quality)
OpenAI voice API - very good, but censors words
OpenAI realtime voice - only available in the ChatGPT app. The best. Realtime.
What constrains open-source voice?
Voice models are based on transformers.
Transformers are the best next to…
Keep reading with a 7-day free trial
Subscribe to The AI TV Catalogue to keep reading this post and get 7 days of free access to the full post archives.