What question did this study set out to answer?

The aim is to present a new recurrent neural network language model and evaluate its performance in speech recognition tasks.

September 26, 2010

Recurrent neural network based language model

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates improved speech recognition outcomes in language modeling, highlighting RNN advantages.

The aim is to present a new recurrent neural network language model and evaluate its performance in speech recognition tasks.
Developed a recurrent neural network based language model with applications to speech recognition.
Conducted experiments comparing different models including a state of the art backoff language model.
Measured performance using perplexity and word error rate metrics.
Achieved around 50% reduction in perplexity compared to the backoff language model.
Demonstrated an 18% reduction in word error rate on the Wall Street Journal task.
Observed a 5% reduction in error rate on the NIST RT05 task, even when the backoff model had more training data.

Mikolov et al. (2010) studied this question.

View Full Paper

Ask AI

Bookmark