Hume says its Audio Decision API answers questions about audio—such as speaker count, laughter, or caller frustration—in a single call, replacing what previously required more than 10 models. The company states it was trained on over 10 million human judgments and supports both real-time and batch processing, with response times generally between 200ms and 500ms. Pricing is stated at 47 cents per 1M tokens, and the API supports 50+ languages. Access is currently by request.

AI summary of the linked source — not editorial content.

Explore the topic

Loading related coverage…

Explore the full feed

Comments

Loading comments…