
by Hume AI Inc.
Hume's Audio Decisions API lets you ask plain-language questions about an audio file and get back probabilities estimating how human raters would answer.
APIUsage-based
Step 1: Create an account and an API key, choosing its permissions and expiry.
Step 2: Write your questions in plain language and pick a type for each.
Step 3: Send a request with the questions as JSON and one MP3 or WAV file.
Step 4: Read the probabilities in the response, keyed by your question IDs.
Step 5: Apply thresholds or weights in your code to decide what to do.
No. It returns probabilities or time spans for your questions, and your code makes the decision.
MP3 or WAV, 1 second to 5 minutes long, with a request body of at most 25,000,000 bytes. Other formats such as WebM or M4A are rejected.
Not directly. API keys must not ship in client-side code, so the browser sends audio to your server, which forwards it to Hume.
It is HTTP only with one file per request, so for live audio you send clips as they are captured, at least 1 second each.
Rime TTSRime TTS applies AI to low-latency text-to-speech API designed for natural conversational voice agents and production speech applications.
WellSaid APIWellSaid API applies AI to enterprise AI voice API for generating consistent branded narration and spoken content from text.
RimeRime is a speech platform with low-latency text-to-speech, speech-to-text and a speech-to-speech API for voice agents. It is used through HTTP, WebSocket and gRPC APIs, a CLI and an MCP server.