Lipreading with Long Short-Term Memory

Abstract

Lipreading, i.e. speech recognition from visual-only recordings of a speaker's face, can be achieved with a processing pipeline based solely on neural networks, yielding significantly better accuracy than conventional methods. Feed-forward and recurrent neural network layers (namely Long Short-Term Memory; LSTM) are stacked to form a single structure which is trained by back-propagating error gradients through all the layers. The performance of such a stacked network was experimentally evaluated and compared to a standard Support Vector Machine classifier using conventional computer vision features (Eigenlips and Histograms of Oriented Gradients). The evaluation was performed on data from 19 speakers of the publicly available GRID corpus. With 51 different words to classify, we report a best word accuracy on held-out evaluation speakers of 79.6% using the end-to-end neural network-based solution (11.6% improvement over the best feature-based solution evaluated).

Accepted for publication at ICASSP 2016

References in corpus (1)

• LSTM: A Search Space Odyssey

Cited by in corpus (11)

• Deep Audio-Visual Speech Recognition
• Phased LSTM: Accelerating Recurrent Network Training for Long or Event-based Sequences
• LipNet: End-to-End Sentence-level Lipreading
• SottoVoce: An Ultrasound Imaging-Based Silent Speech Interaction Using Deep Neural Networks
• Phoneme-to-viseme mappings: the good, the bad, and the ugly
• Learning Spatio-Temporal Features with Two-Stream Deep 3D CNNs for Lipreading
• Vid2speech: Speech Reconstruction from Silent Video
• Resource aware design of a deep convolutional-recurrent neural network for speech recognition through audio-visual sensor fusion
• Modality Attention for End-to-End Audio-visual Speech Recognition
• Towards Pose-invariant Lip-Reading
• The speaker-independent lipreading play-off; a survey of lipreading machines