Data selection from multiple ASR systems' hypotheses for unsupervised acoustic model training.
Sheng LiYuya AkitaTatsuya KawaharaPublished in: ICASSP (2016)
Keyphrases
- data sets
- high quality
- data analysis
- computer systems
- data collection
- knowledge discovery
- image data
- training set
- data sources
- data processing
- small number
- original data
- raw data
- spatial data
- supervised learning
- database
- data points
- relational databases
- data structure
- training data
- distributed systems
- input data
- hidden markov models
- experimental data
- data acquisition
- data mining
- multiple sources
- training dataset
- storage systems