Multi-microphone Automatic Speech Segmentation in Meetings Based on Circular Harmonics Features.
Théo MariotteAnthony LarcherSilvio MontrésorJean-Hugh ThomasPublished in: INTERSPEECH (2023)
Keyphrases
- speaker diarization
- features extraction
- fully automatic
- feature extraction
- low level
- segmentation algorithm
- segmentation accuracy
- feature vectors
- automatic speech recognition
- medical images
- level set
- frequency domain
- speech recognition
- global shape
- audio visual
- extracted features
- text to speech
- fundamental frequency
- lexical features
- grey level
- object segmentation
- shape prior
- segmentation method
- graph cuts
- feature set
- input image
- image features
- feature space