Improving RNN-T ASR Performance with Date-Time and Location Awareness
This work addresses performance enhancement in ASR for virtual assistants, showing incremental gains through context integration.
The paper tackled improving automatic speech recognition for virtual assistants by incorporating date-time and location context into an RNN-T model, resulting in up to 4.62% relative improvement overall and up to 11.5% in specific domains without significant degradation elsewhere.
In this paper, we explore the benefits of incorporating context into a Recurrent Neural Network (RNN-T) based Automatic Speech Recognition (ASR) model to improve the speech recognition for virtual assistants. Specifically, we use meta information extracted from the time at which the utterance is spoken and the approximate location information to make ASR context aware. We show that these contextual information, when used individually, improves overall performance by as much as 3.48% relative to the baseline and when the contexts are combined, the model learns complementary features and the recognition improves by 4.62%. On specific domains, these contextual signals show improvements as high as 11.5%, without any significant degradation on others. We ran experiments with models trained on data of sizes 30K hours and 10K hours. We show that the scale of improvement with the 10K hours dataset is much higher than the one obtained with 30K hours dataset. Our results indicate that with limited data to train the ASR model, contextual signals can improve the performance significantly.