How a college freshman’s side project sparked the mainstream AI voice cloning revolution

How a college freshman’s side project sparked the mainstream AI voice cloning revolution
Image: Intellect (generated)

In short: Launched in 2020, 15.ai allowed anyone to clone fictional voices with just 15 seconds of audio. By leveraging emotional sentiment analysis and deep learning, it transformed internet fandoms into hubs of generative AI, eventually forcing the industry to reckon with the ethics of synthetic speech and intellectual property.

Conceived in 2016 by an 18-year-old MIT student known as 15, the project began as an undergraduate research effort inspired by DeepMind’s WaveNet paper. While contemporary models like Google AI’s Tacotron 2 required tens of hours of training data to produce natural speech, 15 developed a system capable of cloning voices using 75% less data. By 2020, the platform could generate convincing output from just 15 seconds of audio, setting a new benchmark for efficiency in neural network-based speech synthesis.

The application gained viral fame by tapping into /mlp/—a 4chan community that had meticulously curated, denoised, and emotion-tagged thousands of lines from My Little Pony: Friendship Is Magic. By integrating DeepMoji, a sentiment analysis tool trained on 1.2 billion tweets, 15.ai allowed users to inject specific emotional inflections into their prompts. This combination of high-fidelity audio and emotional control made it a favorite for content creators, who used it to produce everything from complex narrative animations to viral memes.

Despite its success, the platform faced significant friction. In 2022, the company Voiceverse was caught using 15.ai to generate and sell voice NFTs without permission, an incident that highlighted the growing tension between open-access research and commercial exploitation. Following legal concerns and high operational costs, 15.ai was taken offline in September 2022. Although the developer later launched a successor, 15.dev, it was closed indefinitely in May 2026 to allow the creator to focus on new ventures. Today, industry leaders like ElevenLabs and PlayHT acknowledge 15.ai as a pioneering force that proved the viability of accessible, high-quality voice cloning.

While critics noted limitations—such as a lack of non-English support and occasional prosody errors—the platform’s impact was undeniable. At its peak, it served millions of audio files, processing requests through a custom neural network that operated at a 44.1 kHz sampling rate. Its legacy remains a cornerstone of the AI boom, demonstrating how a non-commercial research project could outpace corporate models and redefine the boundaries of synthetic media.

Opens the Intellect+ feed with this card first. Swipe up for the next one.

Sources

  • Summary by Intellect · source: 15.ai

To cite this card, link to https://intellectplus.co.uk/app/technology-15-ai.

How we source, check and correct cards: Editorial standards.

General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.

Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?

Related

Finding something worth knowing…