Achieving Human Parity in Conversational Speech Recognition
This work solves the long-standing challenge of matching human performance in speech recognition for conversational tasks, marking a milestone in the field.
The paper tackled the problem of conversational speech recognition by measuring human error rates on the NIST 2000 test set and demonstrated that their automated system achieved human parity, with error rates of 5.8% on Switchboard and 11.0% on CallHome, slightly outperforming human benchmarks of 5.9% and 11.3%, respectively.
Conversational speech recognition has served as a flagship speech recognition task since the release of the Switchboard corpus in the 1990s. In this paper, we measure the human error rate on the widely used NIST 2000 test set, and find that our latest automated system has reached human parity. The error rate of professional transcribers is 5.9% for the Switchboard portion of the data, in which newly acquainted pairs of people discuss an assigned topic, and 11.3% for the CallHome portion where friends and family members have open-ended conversations. In both cases, our automated system establishes a new state of the art, and edges past the human benchmark, achieving error rates of 5.8% and 11.0%, respectively. The key to our system's performance is the use of various convolutional and LSTM acoustic model architectures, combined with a novel spatial smoothing method and lattice-free MMI acoustic training, multiple recurrent neural network language modeling approaches, and a systematic use of system combination.