Drama 3 ranks high on Artificial Analysis speech speed, price and pronunciation tests
Overview
Artificial Analysis reports that Drama 3 generates speech at 66.6 characters per second for $15 per million characters.
That is faster than Drama 3 Preview at 60.7 and Murf's Speech Gen 2 at 63.0, but slower than ElevenLabs' Eleven v4 at 70.0. It is cheaper than Gemini 3.8 Flash TTS at $16.5 per million and Eleven v4 at $80 per million, and Artificial Analysis says every model ranked above it on Provider Voice costs more.
In a separate post, Artificial Analysis says Drama 3 scores 94.6% overall on its Pronunciation Robustness benchmark, first among all models it tested, open weights or proprietary. Human reviewers judged whether models pronounce challenging text correctly. Drama 3 ranks first in Expanding Shorthand at 96.6% and second in Contextually Appropriate at 96.5%, but 13th in Standalone Terms at 91.8%.
Written by AI from the articles below · updated Oct 29, 11:02 AM ET
Check the sources:
Developments
2 developments
- Oct 29, 11:00 AM ET · 1 articleArtificial Analysis ranks Drama 3 on speech generation speed and priceArtificial Analysis: Drama 3 generates speech at 66.6 characters per second for $15 per million
- Oct 29, 11:00 AM ET · 1 articleArtificial Analysis ranks Drama 3 first on its Pronunciation Robustness benchmarkArtificial Analysis: Drama 3 tops Artificial Analysis Pronunciation Robustness benchmark at 94.6%
Article timeline
The articles in this story. Times are ET.
Artificial Analysis@ArtificialAnlysOfficialDrama 3 generates speech at 66.6 characters per second for $15 per millionAIDrama 3 generates 66.6 characters per second, ahead of Drama 3 Preview at 60.7 and Murf's Speech Gen 2 at 63.0, but behind ElevenLabs' Eleven v4 at 70.0. Drama 3 costs $15 per million characters, below Gemini 3.8 Flash TTS at $16.5 and Eleven v4 at $80, and every model ranked above it on Provider Voice costs more.

Artificial Analysis@ArtificialAnlysOfficialDrama 3 tops Artificial Analysis Pronunciation Robustness benchmark at 94.6%AIDrama 3 scores 94.6% overall on the Artificial Analysis Pronunciation Robustness benchmark, ranking first among all models tested, open weights or proprietary. The benchmark has human reviewers judge whether text-to-speech models pronounce challenging text correctly across four categories. Drama 3 ranks first in Expanding Shorthand at 96.6%, second in Contextually Appropriate at 96.5%, third in Preserving Exact Sequences at 87.1%, and 13th in Standalone Terms at 91.8%.

Heat Trend
Not enough continuous observations to show a trend yet.