This technology provides multiple speech identification by identifying sound in a way that uniquely mimics the ear-to-brain interconnection through human voice identification learning and recognition training. The object is real-time or faster voice identification needing only relatively simple computing resources. This invention looks for prosody matches (spectral patterns over time periods that were trained into a software Artificial Neural Network (ANN)–based model). The Infinite Impulse Response (IIR) filter patent can be singularly applied to other uses as well.
Potential applications include: voice recognition for home automation, sound detections for security systems, assistive technology, and audio forensics.
- An IIR filter provides for computational efficiency and meaningful immediate output following a one-cycle initialization - no staging delays, when given an input
- IIR resonator digital filter can be utilized in a biomimetic voice identifier
- US patents 8,442,825 and 8,396,912 available for license
- Potential for collaboration with NSA scientists and engineers