On-Device Speech Recognition for Mobile Devices and Custom Hardware
summary
Company claims a working KeenASR SDK for on-device speech recognition that runs on iOS, Android and Linux and has been developed for about five years, built as a bootstrap project. Demonstrations and customers include EdTech deployments, a voice-controlled metronome and enterprise field tools. Key gains are privacy by design (COPPA), offline resilience, lower and predictable costs, and low latency. A downloadable 15-minute trial SDK lets teams evaluate models and periodic app-delivered updates.
AI-written summary · about
chapters (9)
- 0:00Introduction to talk
- 0:41Speaker background and journey
- 4:37Origin of on-device speech recognition
- 6:03Benefits of on-device processing
- 9:16Use cases and market traction
- 12:49Future growth in speech tech
- 15:17Privacy vs connectivity importance
- 17:21Handling on-device updates
- 19:04How to reach Keen Research
related talks
36mVoice Enabling Mobile Apps and Custom Hardware Devices
06mSensing to Hear: Speech Enhancement for Mobile Devices Using Acoustic Signals
57mApplied AI and Accessibility to Play Games in New Ways
34mVoice Search Using NLP & CoreML
35mHaving a Voice on Android: How hard can it be?
44mProject CapisciAMe
20mLip-Interact: Improving Mobile Device Interaction with Silent Speech Commands
56mSpeech Recognition: How to Build Speech-to-Text Yourself