Hume AI

Hume AI is an Empathic AI Research Lab focused on embedding emotional intelligence into voice models. They provide open source models, datasets, and evaluation APIs that enhance the capabilities of voice AI. Their decades of research enable speakers to reflect a wide range of emotional expressions and utilize high-quality voice descriptors.

Hume AI has curated a library of speech datasets covering various domains such as gaming, health, and business, facilitating the development of sophisticated voice AI systems. The platform also offers a Human Feedback API, simplifying the process of gathering reliable feedback from vetted participants. This ensures that models are tested for listenability, audio quality, and smoothness.

The emphasis on high-quality, annotated speech datasets and human evaluation underpins Hume AI's mission to improve voice model performance across multiple languages and emotional contexts. Their open-source LLM TTS system, known as TADA, is designed to streamline text and audio processing while minimizing common challenges like latency.