Huggingface

Hurston-1 is an open-source narrowband text-to-speech (TTS) model released under the Apache-2.0 license by Destined AI, an Atlanta-based company. The developer reports strong performance on American English, with particular strength in Southern American and Regional African English accents.

The model is hosted on Hugging Face. It is built on voice data collected since 2021 from over 20,000 people, with fully consented recordings totaling about one million audio clips. The stated goal is to help more people feel comfortable using and interacting with AI.