Hume AI

Hume AI provides open source models, datasets, and evaluation APIs to embed emotional intelligence into voice AI systems. Their offerings include:

• Human Feedback API: Run human evaluation studies with science-backed survey templates to gather preference feedback on voice models.
• Data Library: Curated speech datasets covering conversational audio, emotional reproduction, multilingual audio, voice realism, domain-specific, and task-specific data across 50+ languages and 48 emotions.
• Models:
  • TADA (open source): LLM TTS system that streams text and audio together to reduce hallucinations and latency.
  • Octave (closed source): LLM TTS with voice design, modulation, cloning, and conversion.
  • EVI (closed source): LLM Speech-to-Speech system with interruptibility, back channeling, and expressive instruction following.

Key Benefits

• Open source models and datasets available
• Supports 50+ languages and 48 emotions
• Provides evaluation APIs for human feedback
• Offers both open and closed source models