Dear all,
This is a reminder of today's tea talk by Jan Chorowski.
Hope to see many of you there! - K
On Fri, Oct 3, 2014 at 1:37 PM, Kyung Hyun Cho cho.k.hyun@gmail.com wrote:
Dear all,
On Wednesday (8 Oct), Jan Chorowski will tell us about his latest work on building an end-to-end speech recognition system using attention mechanism. This work has been submitted to NIPS Deep Learning Workshop 2014, and the submitted manuscript is attached.
Hope to see many of you there!
- K
===
- Title: End-to-end Continuous Speech Recognition using Attention-based
Recurrent NN: First Results
- Spaker: Prof. Jan Chorowski, University of Wrocław, Poland
- Date and Place: 13.30-14.30, 8 Oct (Wednesday) @AA3195
- Abstract:
We replace the Hidden Markov Model (HMM) which is traditionally used in in continuous speech recognition with a bi-directional recurrent neural network en- coder coupled to a recurrent neural network decoder that directly emits a stream of phonemes. The alignment between the input and output sequences is established using an attention mechanism: the decoder emits each symbol based on a context created with a subset of input symbols selected by the attention mechanism. We report initial results demonstrating that this new approach achieves phoneme error rates that are comparable to the state-of-the-art HMM-based decoders, on the TIMIT dataset.