# Hume launches an audio decision API that answers questions from raw speech

> Original title: A lot of the meaning in speech sits in the delivery, and models can now be asked about it directly.

- Source: Rohan Paul
- Published: 2026-11-02T16:56:38.000Z
- Bad Influence: https://badinfluence.news/items/etsec15s0bx372xptjlfxzxdz
- Original: https://x.com/rohanpaul_ai/status/2117299858243735883

## Summary

Hume has introduced the Hume Audio Decision API, which answers questions about a clip directly from the audio without a transcript step. Developers can ask questions such as whether a caller is angry and receive a direct answer with a confidence percentage. The company says the API is trained on more than 10 million human judgments and can run in realtime or across hours of recordings.

## Why it matters

The source shows a voice-based API that answers questions from raw audio, a contrast with transcript-based systems that can miss tone and sarcasm.

## Post

A lot of the meaning in speech sits in the delivery, and models can now be asked about it directly.

The Hume Audio Decision API answers questions about a clip from the audio itself, with no transcript step, and returns a probability for each answer you define.

> puts sighs, laughter, hesitation, overlapping speech and background sounds inside the model’s input. A transcript would have reduced all of them to the same words.

> so you can ask like “Is this caller angry?”, and get a direct answer with a confidence percentage.

> Most AI judges calls by reading a text transcript, which loses tone; But this one listens to the voice itself, so a sarcastic “Oh, perfect” no longer counts as happy.

https://x.com/hume_ai/status/2117285608745820372/video/1

[Video](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Famplify_video_thumb%2F2117276914591399937%2Fimg%2FUT3OrHoJmR0zyCbG.jpg&mode=thumb&exp=1793836800&sig=c73c50110da6b465)

## Quoted post @hume_ai

> Introducing the Hume Audio Decision API.
> 
> Ask anything: How many speakers? Is anyone laughing? Is the caller frustrated?
> 
> What used to take 10+ models is now one call.
> 
> Trained on 10M+ human judgments, so it answers the way human listeners would.
> 
> Realtime, or across hours of recordings.
> 
> Request access below.

https://x.com/hume_ai/status/2117285608745820372