CIF-RNNT: Streaming ASR Via Acoustic Word Embeddings with Continuous Integrate-and-Fire and RNN-Transducers.

Wen Shen Teo Yasuhiro Minami

Published in: ICASSP (2024)

Keyphrases

automatic speech recognition
speech recognizers
speech recognition
speech recognition systems
recurrent neural networks
spontaneous speech
prosodic features
speech sounds
continuous query processing
nearest neighbor

co occurrence
word recognition
data streams
recognition errors
real time
finite automata
information retrieval
word error rate
spoken language

acoustic features
streaming data
dimensionality reduction
speech recognizer
n gram
video streaming
vector space
word level
spoken document retrieval

continuous functions
source localization
distance measure
binary codes
conversational speech
speech signal
acoustic signal
keywords
multimedia