Cartesia's Sonic-3.5 Takes #1 on Artificial Analysis Speech Leaderboard (artificialanalysis.ai)

Cartesia's Sonic-3.5 has taken the top spot on the Speech Arena's Text to Speech Leaderboard, achieving an impressive Elo score of 1218. This ranking is determined through blind comparisons where listeners evaluate pairs of speech samples to decide which sounds more natural, underscoring the model's ability to produce high-quality, lifelike speech. Sonic-3.5 outperformed other notable models, including Gemini 3.1 Flash TTS and Realtime TTS 1.5 Max, positioning itself as a leader in the competitive field of text-to-speech technology. The significance of Sonic-3.5's achievement for the AI/ML community lies in its validation of advanced speech synthesis techniques and user-centered design. Higher Elo scores indicate not only technical excellence but also greater acceptance from users, highlighting the importance of natural-sounding speech in applications like virtual assistants, entertainment, and customer service. Additionally, the leaderboard showcases various models, including Fish Audio S2 Pro as the highest-ranked open weights model, emphasizing the diverse landscape of AI-driven speech technologies and the importance of accessibility and affordability in this evolving field.