English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

There is often a trade-off between performance and latency in streamingautomatic speech recognition (ASR). Traditional methods such as look-ahead andchunk-based methods, usually require information from future frames to advancerecognition accuracy, which incurs inevitable latency even if the computationis fast enough. A causal model that computes without any future frames canavoid this latency, but its performance is significantly worse than traditionalmethods. In this paper, we propose corresponding revision strategies to improvethe causal model. Firstly, we introduce a real-time encoder states revisionstrategy to modify previous states. Encoder forward computation starts once thedata is received and revises the previous encoder states after several frames,which is no need to wait for any right context. Furthermore, a CTC spikeposition alignment decoding algorithm is designed to reduce time costs broughtby the revision strategy. Experiments are all conducted on Librispeechdatasets. Fine-tuning on the CTC-based wav2vec2.0 model, our best method canachieve 3.7/9.2 WERs on test-clean/other sets, which is also competitive withthe chunk-based methods and the knowledge distillation methods.

Improving Streaming End-to-End ASR on Transformer-based Causal Models with Encoder States Revision Strategies