Model-based speech/non-speech segmentation of a heterogeneous multilingual TV broadcast collection.
Brecht DesplanquesJean-Pierre MartensPublished in: ISPACS (2013)
Keyphrases
- speech recognition
- speech signal
- multi lingual
- fully unsupervised
- audio visual
- automatic speech recognition
- tv broadcast
- speech synthesis
- image segmentation
- level set
- dialogue system
- image analysis
- segmentation method
- region growing
- medical images
- emotion recognition
- text to speech
- noisy environments
- recognition engine
- spontaneous speech
- multiscale
- digital libraries
- spoken language
- contextual information
- hidden markov models
- image data
- object segmentation
- document collections