Hume Audio Decision API

API that answers plain-language questions about one audio file and returns probabilities.

Also known as Audio Decision API, Hume Audio Decisions

What it does

Hume's Audio Decision API, announced November 2, 2026, takes one audio file and plain-language questions and returns probabilities. The blog describes yes-or-no, choice, scale, and time-span answers. The changelog says each question uses one of four primitives, noul, choice, score, or span, and that one request answers every question about that file. The limits page says a request takes one MP3 or WAV file of 5 seconds to 5 minutes, up to 25,000,000 bytes, up to 20 noul, choice, and score questions, and up to 5 span questions, and that each organization can send 60 requests per minute. The billing page prices a successful request at $0.47 per million input tokens. Audio, questions, and context are billed. Answers are not. A request rejected with a 4xx status is not billed. An organization gets a trial credit of 1 million input tokens once it is granted access. The blog says an initial evaluation reached 70.7% agreement with a human panel's average judgment, compared with 56.5% for the individual-rater baseline. That comparison is Hume's. The blog itself does not print a price.

Who is Hume Audio Decision API for?

Developers who need a probability for a question about speech or sound, instead of a transcript.

Tags

• api
• speech
• decisions