H EVAL : A New Hybrid Evaluation Metric for Automatic Speech Recognition Tasks
Zitha Sasindran, Harsha Yelchuri, T. V. Prabhakar, Supreeth Rao · 2023
Many studies have examined the shortcomings of word error rate (WER) as an evaluation metric for automatic speech recognition (ASR) systems. Since WER considers only literal word-level correctness, new evaluation metrics based on semantic similarity such as semantic distance (SD) and BERTScore have been developed. However, we found that these metrics have their own limitations, such as a tendency to overly prioritise keywords. We propose HEVAL, a new hybrid evaluation metric for ASR systems that considers both semantic correctness and error rate and performs significantly well in scenarios where WER and SD perform poorly. Due to lighter computation compared to BERTScore, it offers 49 times reduction in metric computation time. Furthermore, we show that HEVALcorrelates strongly with downstream NLP tasks. Also, to reduce the metric calculation time, we built multiple fast and lightweight models using distillation techniques without significant reduction in performance.