Evaluating the reliability of acoustic speech embeddings.
Robin AlgayresMohamed Salah ZaïemBenoît SagotEmmanuel DupouxPublished in: CoRR (2020)
Keyphrases
- speech sounds
- speech recognition systems
- speech recognition
- acoustic features
- prosodic features
- speech signal
- vector space
- emotional speech
- automatic speech recognition
- speech synthesis
- low dimensional
- acoustic signal
- audio visual
- spoken language
- speech recognizers
- speaker independent
- dimensionality reduction
- vocal tract
- endpoint detection
- text to speech
- multi modal
- high dimensional data
- source localization
- formant frequencies
- failure rate
- acoustic models
- manifold learning
- speech segments
- vowel phonemes
- reliability analysis
- information retrieval
- noisy speech
- euclidean space
- emotion recognition
- bird species
- audio features
- underwater acoustic
- highly reliable