Toward Zero Oracle Word Error Rate on the Switchboard Benchmark
This work addresses evaluation inconsistencies in ASR for researchers and practitioners, but it is incremental as it focuses on methodological refinements rather than new models.
The paper tackled the Switchboard benchmark's practical evaluation issues by correcting reference transcriptions and altering scoring, achieving a record 2.3% WER for research systems and proposing a more discriminating metric. It also explored oracle WER methods, reaching 0.18% using dense lattices.
The "Switchboard benchmark" is a very well-known test set in automatic speech recognition (ASR) research, establishing record-setting performance for systems that claim human-level transcription accuracy. This work highlights lesser-known practical considerations of this evaluation, demonstrating major improvements in word error rate (WER) by correcting the reference transcriptions and deviating from the official scoring methodology. In this more detailed and reproducible scheme, even commercial ASR systems can score below 5% WER and the established record for a research system is lowered to 2.3%. An alternative metric of transcript precision is proposed, which does not penalize deletions and appears to be more discriminating for human vs. machine performance. While commercial ASR systems are still below this threshold, a research system is shown to clearly surpass the accuracy of commercial human speech recognition. This work also explores using standardized scoring tools to compute oracle WER by selecting the best among a list of alternatives. A phrase alternatives representation is compared to utterance-level N-best lists and word-level data structures; using dense lattices and adding out-of-vocabulary words, this achieves an oracle WER of 0.18%.