A recent innovation in speech technology from SRI International's Speech Technology & Research (STAR) Laboratory is SenSay Analytics™ , a platform that performs real-time speaker state classification from spoken audio. SRI is working with both research and industry partners using SenSay Analytics to estimate speaker state—including emotion, sentiment, cognition, health, mental health and communication quality—in a range of end applications.
At sub-second intervals, the platform updates both features and class estimates using advanced signal features that capture spectral, prosodic, articulatory, auditory, discourse and fluency characteristics, as well as features designed specifically for robustness to noise and reverberation. The platform can analyze the features from the signal alone or combined with automatic speech recognition to model word-based information via sentiment models. Features are modeled using state-of-the-art machine learning approaches appropriate to the task, training data and application constraints.
Provides class and feature updates at < 1 second; crucial for applications such as driver monitoring, dialog system response and customer service.
Deploy on premises or in the cloud. Use as a feature extractor, class predictor or both. Use with or without automatic speech recognition.
Can be adapted to task, domain, language, single- or multi-party conversations.
Can run in cloud, on laptop/desktop or in client-hosted environment.
APIs let clients add additional sensor capabilities such as video-based or physiological features.
Architected to support simultaneous live streams.