Hugging Face launches an open-source TTS leaderboard

On September 30, 2026, Hugging Face published an open-source TTS leaderboard on its official blog. That post is so far the only report on the launch, dated 08:00 the same day.

According to the post, the leaderboard covers multilingual text-to-speech and voice cloning, and offers scalable evaluation, with the stated goal of making speech model evaluation more transparent. Those are the limits of what the report confirms: the publisher is Hugging Face, the artifact is an open-source leaderboard, the capabilities it addresses are multilingual TTS and voice cloning, the evaluation is described as scalable, and the aim is transparency in speech model assessment.

The post does not say how many models the leaderboard includes, who submitted them, which datasets or benchmarks are used, or whether any rankings, scores, or comparisons with other leaderboards exist. No third-party organization or vendor is quoted. The story therefore stands where it started: Hugging Face has released an open-source TTS leaderboard, and no further details or follow-up developments appear in the available material.