Dissecting User-Perceived Latency of On-Device E2E Speech Recognition

by   Yuan Shangguan, et al.

As speech-enabled devices such as smartphones and smart speakers become increasingly ubiquitous, there is growing interest in building automatic speech recognition (ASR) systems that can run directly on-device; end-to-end (E2E) speech recognition models such as recurrent neural network transducers and their variants have recently emerged as prime candidates for this task. Apart from being accurate and compact, such systems need to decode speech with low user-perceived latency (UPL), producing words as soon as they are spoken. This work examines the impact of various techniques – model architectures, training criteria, decoding hyperparameters, and endpointer parameters – on UPL. Our analyses suggest that measures of model size (parameters, input chunk sizes), or measures of computation (e.g., FLOPS, RTF) that reflect the model's ability to process input frames are not always strongly correlated with observed UPL. Thus, conventional algorithmic latency measurements might be inadequate in accurately capturing latency observed when models are deployed on embedded devices. Instead, we find that factors affecting token emission latency, and endpointing behavior significantly impact on UPL. We achieve the best trade-off between latency and word error rate when performing ASR jointly with endpointing, and using the recently proposed alignment regularization.


page 1

page 2

page 3

page 4


Streaming End-to-end Speech Recognition For Mobile Devices

End-to-end (E2E) models, which directly predict output character sequenc...

Training Neural Speech Recognition Systems with Synthetic Speech Augmentation

Building an accurate automatic speech recognition (ASR) system requires ...

Two-Pass End-to-End Speech Recognition

The requirements for many applications of state-of-the-art speech recogn...

Flexi-Transducer: Optimizing Latency, Accuracy and Compute forMulti-Domain On-Device Scenarios

Often, the storage and computational constraints of embeddeddevices dema...

Low Latency ASR for Simultaneous Speech Translation

User studies have shown that reducing the latency of our simultaneous le...

Scaling Up Online Speech Recognition Using ConvNets

We design an online end-to-end speech recognition system based on Time-D...

Interactive decoding of words from visual speech recognition models

This work describes an interactive decoding method to improve the perfor...