Qualitative Evaluation of Language Model Rescoring in Automatic Speech Recognition
For ASR researchers and practitioners, this work introduces more informative evaluation metrics, but it is an incremental contribution as it applies existing NLP metrics to a known problem.
The paper proposes two new metrics, POSER and EmbER, to evaluate the linguistic quality of ASR transcriptions beyond word error rate, and uses them to analyze the impact of language model rescoring. Results show that these metrics provide insights into grammatical and semantic improvements not captured by WER alone.
Evaluating automatic speech recognition (ASR) systems is a classical but difficult and still open problem, which often boils down to focusing only on the word error rate (WER). However, this metric suffers from many limitations and does not allow an in-depth analysis of automatic transcription errors. In this paper, we propose to study and understand the impact of rescoring using language models in ASR systems by means of several metrics often used in other natural language processing (NLP) tasks in addition to the WER. In particular, we introduce two measures related to morpho-syntactic and semantic aspects of transcribed words: 1) the POSER (Part-of-speech Error Rate), which should highlight the grammatical aspects, and 2) the EmbER (Embedding Error Rate), a measurement that modifies the WER by providing a weighting according to the semantic distance of the wrongly transcribed words. These metrics illustrate the linguistic contributions of the language models that are applied during a posterior rescoring step on transcription hypotheses.