Fish Audio takes on ElevenLabs with ultra-cheap S1 model

ElevenLabs has long been the gold standard for AI voice generation, but running high-volume audio on it gets expensive fast. Fish Audio wants to change that math.

Helena at Fish Audio just publicly launched S1, pitched as the most expressive and natural text-to-speech model currently available.

Undercutting the market leader

Fish Audio is taking direct aim at ElevenLabs by offering S1 at one-sixth of the price. Anyone looking to test the waters can clone their own voice for free right now.

Alongside the launch, Helena shared some traction metrics: the platform already counts over 20,000 active developers and has hit $5 million in annual recurring revenue.

Why it matters

For indie hackers and developers building voice agents or churning out podcast clips, voice generation costs can easily kill a project's margins. Bringing ElevenLabs-tier naturalness down by 6x makes high-volume voice apps viable for smaller teams.

We think the AI audio space is officially entering its pricing compression phase. Early leaders enjoyed fat margins on brand reputation alone, but Fish Audio backing up its price war with real developer traction shows that open alternatives and cheaper inference are quickly catching up.