Low-Power Automatic Speech Recognition Through a Mobile GPU and a Viterbi Accelerator.
Reza YazdaniAlbert SeguraJosé-María ArnauAntonio GonzálezPublished in: IEEE Micro (2017)
Keyphrases
- low power
- automatic speech recognition
- hidden markov models
- wireless transmission
- speech recognition
- parallel implementation
- power consumption
- low cost
- high speed
- broadcast news
- speech signal
- conversational speech
- word error rate
- single chip
- digital signal processing
- real time
- vlsi architecture
- mobile devices
- spoken words
- recognition errors
- noisy environments
- viterbi algorithm
- speech retrieval
- low power consumption
- vlsi circuits
- speech corpus
- ultra low power
- logic circuits
- parallel computing
- spontaneous speech
- power reduction
- mixed signal
- parallel processing
- signal processing
- feature extraction
- cmos technology
- context aware
- multiscale
- image processing
- machine learning