Citrinet: Closing the Gap between Non-Autoregressive and Autoregressive End-to-End Models for Automatic Speech Recognition

TL;DR

Citrinet is a deep convolutional CTC-based speech recognition model that narrows the performance gap between non-autoregressive and autoregressive models, achieving near state-of-the-art accuracy on multiple datasets.

Contribution

The paper introduces Citrinet, a novel deep residual convolutional model with separable convolutions and squeeze-and-excitation, significantly improving non-autoregressive ASR performance.

Findings

Achieves near state-of-the-art accuracy on LibriSpeech and other datasets.

Effectively reduces the gap between non-autoregressive and autoregressive models.

Demonstrates strong performance across multiple multilingual speech datasets.

Abstract

We propose Citrinet - a new end-to-end convolutional Connectionist Temporal Classification (CTC) based automatic speech recognition (ASR) model. Citrinet is deep residual neural model which uses 1D time-channel separable convolutions combined with sub-word encoding and squeeze-and-excitation. The resulting architecture significantly reduces the gap between non-autoregressive and sequence-to-sequence and transducer models. We evaluate Citrinet on LibriSpeech, TED-LIUM2, AISHELL-1 and Multilingual LibriSpeech (MLS) English speech datasets. Citrinet accuracy on these datasets is close to the best autoregressive Transducer models.