Title | ||
---|---|---|
A comparative study of HMM-based approaches for the automatic recognition of perceptually relevant aspects of spontaneous German speech melody |
Abstract | ||
---|---|---|
Three approaches to the speaker independent automaticrecognition of melodic aspects of spontaneousGerman are presented. All systems are based on HiddenMarkov Models. Their input is restricted to thespeech signal from which a feature extraction componentderives eleven prosodic features. No additionalinformation -- as commonly used for prosody recognition-- like word chains, word hypotheses, furthersegmental or lexical prosodic information (e.g. stressplacement) is required. The three... |
Year | Venue | Keywords |
---|---|---|
1999 | EUROSPEECH | feature extraction |
Field | DocType | Citations |
Melody,Prosody,Computer science,Speech recognition,Feature extraction,Speaker recognition,Concatenation,Natural language processing,Artificial intelligence,Hidden Markov model,German | Conference | 2 |
PageRank | References | Authors |
0.40 | 1 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Christel Brindöpke | 1 | 3 | 1.43 |
Gernot A. Fink | 2 | 888 | 82.73 |
Franz Kummert | 3 | 103 | 20.41 |