Accelerating recurrent neural network language model based online speech recognition system

TL;DR

This paper introduces methods to accelerate RNN language models for online speech recognition by using lossy compression and CPU-GPU hybrid computation, achieving over fourfold speed improvements with minimal accuracy loss.

Contribution

It proposes a novel lossy compression technique for history vectors and a hybrid CPU-GPU computation strategy to significantly speed up RNNLM-based speech recognition.

Findings

Speed increased by 1.23 times with minimal accuracy degradation.

Real-time recognition achieved with four times faster processing.

Effective CPU-GPU hybrid parallelization demonstrated.

Abstract

This paper presents methods to accelerate recurrent neural network based language models (RNNLMs) for online speech recognition systems. Firstly, a lossy compression of the past hidden layer outputs (history vector) with caching is introduced in order to reduce the number of LM queries. Next, RNNLM computations are deployed in a CPU-GPU hybrid manner, which computes each layer of the model on a more advantageous platform. The added overhead by data exchanges between CPU and GPU is compensated through a frame-wise batching strategy. The performance of the proposed methods evaluated on LibriSpeech test sets indicates that the reduction in history vector precision improves the average recognition speed by 1.23 times with minimum degradation in accuracy. On the other hand, the CPU-GPU hybrid parallelization enables RNNLM based real-time recognition with a four times improvement in speed.