Pitchln: eavesdropping via intelligible speech reconstruction using non-acoustic sensor fusion.
Jun HanAlbert Jin ChungPatrick TaguePublished in: IPSN (2017)
Keyphrases
- sensor fusion
- mobile robot
- speech recognition systems
- multi sensor
- acoustic features
- speech recognition
- speech sounds
- monitoring system
- emotional speech
- three dimensional
- audio visual
- prosodic features
- speaker independent
- speech signal
- real time
- multiple sensors
- speech synthesis
- spoken language
- high resolution
- automatic speech recognition
- cscw systems
- acoustic signal
- text to speech
- speaker verification
- reconstruction method
- image fusion
- acoustic models
- vocal tract
- inertial sensors
- formant frequencies
- vowel phonemes