Pre-training of Speaker Embeddings for Low-latency Speaker Change Detection in Broadcast News.
Leda SariSamuel ThomasMark Hasegawa-JohnsonMichael PichenyPublished in: ICASSP (2019)
Keyphrases
- change detection
- broadcast news
- speaker diarization
- automatic speech recognition
- acoustic models
- low latency
- speaker identification
- speech recognition
- remote sensing
- remote sensing images
- data streams
- hidden markov models
- video retrieval
- highly efficient
- concept drift
- real time
- speech signal
- high speed
- high throughput
- stream processing
- virtual machine
- language processing
- image registration
- machine learning