Interhuman

Interhuman: Social Signal Models for Voice and Video

Interhuman builds AI models that read the social signals in a conversation, such as confidence, hesitation or frustration, rather than only the words in a transcript. Its Inter-2 Audio model detects these signals from the sound of speech alone, while Inter-2 processes voice, face and body language from a single video stream, all served through an API.

Key Features

• Inter-2 Audio: Detects social signals from audio alone, with no camera required, using intonation, emphasis, rhythm and timing.
• Six audio signals: Identifies confidence, interest, agreement, hesitation, uncertainty and frustration in speech.
• Engagement tracking: Classifies each moment as disengaged, neutral or engaged.
• Conversation quality score: A 0-100 score covering clarity, authority, energy, rapport and learning.
• Inter-2 multimodal model: Detects 12 social signals from voice, face and body language in one video stream.
• Published benchmarks: Interhuman reports 82.72% engagement accuracy on 839 held-out audio clips for Inter-2 Audio, measured by the company.
• Developer API: API key access with documentation, a platform dashboard and code examples, plus open demo projects on GitHub.
• Behavioral science grounding: Models are built on research into observable communication patterns.
• Free tier: Start with a free plan that includes the conversation quality score.

Use Cases

• Sales coaching: Spot hesitation, interest and agreement in sales calls to coach reps on what moved the conversation.
• Voice agents and customer service: Give a voice agent a read of the caller's frustration or uncertainty so it can adapt its response.
• Research interviews: Track engagement and confidence across interviews to surface the moments that matter.
• Robotics and meetings: Add social awareness to social robots and remote meeting tools.

Pricing

• Free: EUR 0/month, 900 credits (about 15 minutes), 2 concurrent requests.
• Starter: EUR 59/month, 36,000 credits (about 600 minutes), 5 concurrent requests.
• Scale: EUR 300/month, 240,000 credits (about 4,000 minutes), 30 concurrent requests.
• Enterprise: Custom volume, dedicated support and SLA guarantees.

Getting Started

• Website: https://www.interhuman.ai/
• Documentation: https://docs.interhuman.ai/

Interhuman adds a layer of social understanding to voice and video products, turning how something was said into structured signals that developers can act on.