Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

[Tensorflow] An Improved Deep-Learning Model of Turn-taking in Spoken Dialogue

Many turn-taking models have been built to solve specific tasks, like predicting if a user will hold his/her turn after a pause. However, Skantze showed that a continuous, general turn-taking model can handle multiple tasks successfully. Building on his model, a Long Short-Term Memory (LSTM) Recurrent Neural Network, we made four improvements: we 1) replaced the Sigmoid Units with Parametric Rectified Linear Units, 2) increased the number of layers and neurons, 3) trained without truncation, and 4) added dropout regularization. Our improved model gives good and sometimes better results, up to 26% better. Tensorflow code that builds and trains a general turn-taking model exactly as described by Skantze in his paper 'Towards a General, Continuous Model of Turn-taking in Spoken Dialogue using LSTM Recurrent Neural Networks' ( Tensorflow code that builds and traing a general turn-taking model with the improves described above Code that reads the data used to train either of the two above-mentioned models

You can’t perform that action at this time.