Audio research lab building generalist speech models that follow instructions and learn new tasks in context, accessible via Studio, Live conversation, and API.
Kalpa Labs develops generalist audio models designed to follow instructions, learn new voices or styles in context, and understand nuances of how things are said beyond just speaker identity. The first conversational speech models are in beta through three interfaces: Studio for directing multi-speaker scenes with controlled pacing and emotion, a Live real-time browser-based conversation mode, and a REST API with developer docs and an in-browser playground. Key capabilities include voice cloning, speech continuation, emotional intensity, and steerability. Kalpa Labs is a product of Kalpa Labs.
For people
For agents