AssemblyAI launches AssemblyAI-TTS-1, ranking #12 in Artificial Analysis’s Provider Voice Arena

What happened

AssemblyAI has launched its first Text to Speech model, AssemblyAI-TTS-1, which Artificial Analysis evaluated across its Speech Arenas and Pronunciation Robustness benchmark. It ranks #12 in Provider Voice with an Elo of 1,227 across 1,626 appearances, #10 in Controlled Voice, and #16 in Pronunciation Robustness with 81.7%. It generates 162 characters per second and costs $15/1M characters; only two of the eleven models ranked above it in Provider Voice cost less.

Read next

• More changes in today’s briefingGoogle DeepMind announces Genie 4 for real-time world simulation with native audio

The story so far

AssemblyAI launched its first text-to-speech model, AssemblyAI-TTS-1, and Artificial Analysis ranked it #12 in its Provider Voice Arena, #10 in Controlled Voice and #16 in Pronunciation Robustness. Artificial Analysis reported a Provider Voice Elo score of 1,227 across 1,626 appearances and a Pronunciation Robustness score of 81.7%. The model generates 162 characters per second and costs $15 per million characters; according to Artificial Analysis, only two of the eleven models ranked above it in Provider Voice cost less.