SECP: A Speech Enhancement-Based Curation Pipeline for Scalable Acquisition of Clean Speech.
Adam SabraCyprian M. WronkaMichelle MaoSamer L. HijaziPublished in: ICASSP (2024)
Keyphrases
- speech enhancement
- noisy speech
- noise reduction
- speech signal
- noisy environments
- speech recognition
- signal to noise ratio
- single channel
- hidden markov models
- background noise
- vocal tract
- linear prediction
- additive noise
- smoothing algorithm
- automatic speech recognition
- wiener filter
- speaker identification
- edge detection
- edge preserving
- non stationary
- multi modal
- image compression
- language model
- spectral subtraction