Todd Mozer of Sensory, who have had things up and running throughout the event and way beyond (last 20 years). Embedded Voice Biometrics (small footprints, small calc requirements)–other companies use touch to verify but they’re getting away from that.
Why embedded? Small footprint: lower cost, works efficiently; lower power: less battery replacement, saves $ to consumers. No cloud connection needed: always works, faster responses, lower cost. Can improve accuracy, independent approaches: comparison between client and cloud biometrics, spot checks if/when needed.
Truly handsfree: voice activation, speech recognition and transcription, intent and meaning, data search and query, speech or display response. Voice activation: keyword spotting approach (instead of touch), fast response time and works with high noise. “Hello blue genie” triggers service, even with music in background (demo). Now added: speaker verification, pre-trained phrases (three words, though five words is recommended).
Diverse and growing customer base in consumer electronics including telephony, automotive, home electronics, toys and learning, bluetooth, and more. They had a device that went to Mars. Best selling girls electronics (GirlTech): diary for girls, locked with their voice. Radica acquired them then Mattel acquired them, still best-selling product. Uses 15 year old technology (they offered a free upgrade but was declined).
Increasing accuracy by combining metrics, sensors, platforms, other independent approaches. Increasing client/cloud interactions, including FIDO Alliance, security for transactions, transactions through mobile devices (e.g., “I want to buy that song.”).