Adaptive Endpointing with Deep Contextual Multi-Armed Bandits
Authors:
- 4 June 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Abstract
Current endpointing (EP) solutions learn in a supervised framework, which does not allow the model to incorporate feedback and improve in an online setting. Also, it is common practice to utilize costly grid-search to find the best configuration for an endpointing model. In this paper, we aim to provide a solution for adaptive endpointing by proposing an efficient method for choosing an optimal endpointing configuration given utterance-level audio features in an online setting, while avoiding hyperparameter grid-search. Our method does not require ground truth labels, and uses only online learning from reward signals. Specifically, we propose a deep contextual multi-armed bandit-based approach, combining the representational power of neural networks with the action exploration behavior of Thomp-son modeling algorithms. We compare our approach to several baselines, and show that our deep bandit models also succeed in reducing early cutoff errors while maintaining low latency.Keywords
All Related Versions
This publication has 21 references indexed in Scilit:
- Introducing Deep Reinforcement Learning to Nlu Ranking Tasks2021
- Personalizing Speech Start Point and End Point Detection in ASR Systems from Speaker Embeddings2021
- Personalizing Natural Language Understanding using Multi-armed Bandits and Implicit Feedback2020
- Combining Acoustic Embeddings and Decoding Features for End-of-Utterance Detection in Real-Time Far-Field Speech Recognition Systems2018
- Domain-Specific Utterance End-Point Detection for Speech Recognition2017
- Towards Deep End-of-Turn Prediction for Situated Spoken Dialogue Systems2017
- Turn-Taking Estimation Model Based on Joint Embedding of Lexical and Prosodic Contents2017
- End-of-Utterance Prediction by Prosodic Features and Phrase-Dependency Structure in Spontaneous Japanese Speech2017
- Computationally-efficient endpointing features for natural spoken interaction with personal-assistant systems2014
- A prosody-based approach to end-of-utterance detection that does not require speech recognition2003