Hume Launches Audio Decision API

Hume today launched Audio Decision API, an audio-native model that answers questions about speech and sound. Users choose the questions and the possible answers, anything from whether a voice is artificial intelligence-generated to identifying if the caller is frustrated or if there is a siren in the background noise.

Raw audio goes in, an answer with a confidence score comes out. Users can ask dozens of questions every few seconds on a live call and make an archive with 40,000 audio recordings searchable in minutes.

Questions can cover the following:

• Speech and languages;
• Speakers and voices;
• Emotion and expression;
• Acoustic environment;
• Non-speech events;
• Conversation dynamics;
• Communication quality;
• Intent and outcomes;
• Social and behavioral signals; and
• Temporal and contextual signals.

Teams can also run questions in bulk across audio libraries, track expression over successive segments. and pinpoint the moments when particular sounds or vocal events occur. An audio search engine capability lets users find every call where the customer sounded frustrated.

"For years, Hume has asked panels of listeners to rate emotional expression, naturalness, vocal delivery, and conversational fit. Tens of millions of these judgments have informed Real-World VoiceEQ and helped train our Audio Decision Model to reflect how people perceive speech and sound," the company said in a statement.

In an initial evaluation, Hume's Audio Decision model achieved 67.4 percent agreement with the human panel's average judgment, compared with 54.2 percent for the individual-rater baseline.