On Architectural Issues of Neural Networks in Speech Recognition thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

On Architectural Issues of Neural Networks in Speech Recognition

Published on Jul 31, 20161515 Views

Recently, artificial neural networks (ANN) were able to improve the performance of speech recognition systems dramatically. There have been more than 25 years of extensive research on neural networks

Related categories

Chapter list

On Architectural Issues of Neural Networks in Speech Recognition00:00
Human Language Technology (HLT)00:17
RWTH’s Joint Projects with InterACT: KIT, CMU or HKUST00:53
Evaluation Campaigns: InterACT (KIT)03:25
Statistical Approach: No Alternative (incl. Artificial Neural Networks!)04:39
Hidden Markov Models (HMM) - 105:29
Hidden Markov Models (HMM) - 206:28
Hybrid Approach: HMM and ANN06:59
History: ANN in Acoustic Modelling - 108:00
History: ANN in Acoustic Modelling - 209:45
TDNN: Time Delay Neural Network - 110:45
TDNN: Time Delay Neural Network - 211:45
Today vs. 1988-94: What is Different?13:27
Recurrent Neural Network: String Processing14:12
Direct Model of Label Sequence - 114:22
Direct Model of Label Sequence - 215:06
Comparison with Discriminative/Hybrid HMM15:10
Comparison with CTC: connectionist temporal classification16:04
Direct Model of Label Sequence: Inverted Alignments17:00
Mechanism of Attention: Alignment by ANN17:21
Summary18:23
Congratulations to InterACT and Alex on 25 successful years18:53