GC-TTS: Few-shot Speaker Adaptation with Geometric Constraints.
Ji-Hoon KimSang-Hoon LeeJi-Hyun LeeHonggyu JungSeong-Whan LeePublished in: CoRR (2021)
Keyphrases
- geometric constraints
- speaker adaptation
- text to speech
- speech recognition
- maximum likelihood
- speech synthesis
- automatic speech recognition
- vocal tract
- epipolar geometry
- video shots
- camera calibration
- video sequences
- multiple images
- speech recognizer
- video data
- geometric consistency
- visual features
- video content
- key frames
- speech signal
- low level
- camera motion
- news video
- probabilistic model
- pattern recognition
- information retrieval