Fish Audio, a startup based in Palo Alto, California, raised $50 million in a seed funding round announced on July 28, 2026. Seed funding is the earliest stage of investment in a company. The round was led by Coreline Ventures and Capital Today, with participation from 359 Capital, Parable, Play Time, Alphalist Partners, Bayhouse Ventures, Carya Venture Partners, and HF0. The company was founded by Shijia Liao, a former researcher at NVIDIA, and is led by CEO and co-founder Rissa Cao. TechCrunch

Fish Audio makes artificial intelligence tools that create and clone voices. More than 8 million people use its free and paid tools, and the company brings in $21 million per year from paying customers. Its free software on GitHub, a popular site where developers share code, has received over 31,000 stars, which is a way users show their approval or interest in a project.

In the past year, Fish Audio has released five AI models. Four of them generate spoken audio from text, and one converts spoken audio into text. Three of the voice generation models are free for anyone to download and use. The newest and most advanced one, called S2.1 Pro, is only available through a paid service. The company also sells monthly plans for creators and teams that include voice cloning features.

On the business side, Fish Audio counts HeyGen, Sanas, and Plaud among its customers. The company offers more than 15,000 natural language controls, meaning users can describe how they want the voice to sound using everyday language instead of complicated technical commands. Oskue Honda, a partner at Coreline Ventures, backed the round alongside Capital Today.

The funding comes amid ongoing concerns about voice cloning and consent. Fish Audio faced a controversy in which creators said their voices were uploaded to the platform without their permission. The company responded by automating its takedown process so that unwanted voices are removed in under three minutes. That speeds up the process of taking something down, but it does not address the harder problem of how unauthorized voices get onto the platform in the first place.

The broader context here is the company's strategy of giving away three of its four voice generation models for free while charging for the most advanced one. This is a common approach in the AI industry right now: free versions attract developers and build credibility, while the most powerful tool generates revenue. Whether enough people will keep paying for the advanced version is uncertain, as competing free tools are catching up in quality.

The consent controversy also points to a challenge for the entire voice cloning field, not just Fish Audio. The technology to copy someone's voice from a short audio sample is now easy to use and widely available. Quick takedowns help after the problem occurs, but if regulators eventually require platforms to verify consent before a voice is added, it could significantly change how these businesses operate. Companies that build good consent systems now may be better positioned if those rules arrive.

What the $50 million gives Fish Audio is time and resources to improve its paid models while keeping its free tools available. The customer list, the revenue figure, and the developer interest together suggest the company has moved beyond an experiment into a real business. The remaining question is whether it can grow that business while dealing with the consent issues that come with voice cloning technology. The money and the customers are in place. The rules around permission are not.