No abstract

This paper presents an efficient method to build a corpus to train natural language understanding (NLU) modules. Conventional corpus creation methods involve a common cycle: a subject is given a specific situation where the subject operates a device by voice, and then the subject speaks one utterance to execute the task. In these methods, many subjects are required in order to build a large-scale corpus, which causes a problem of increasing lead time and financial cost. To solve this problem, we propose to incorporate a “probing question” into the cycle. Specifically, after a subject speaks one utterance, the subject is asked to think of alternative utterances to execute the same task. In this way, we obtain many utterances from a small number of subjects. An evaluation of the proposed method applied to interview-based corpus creation shows that the proposed method reduces the number of subjects by 41% while maintaining morphological diversity in a corpus and morphological coverage for user utterances spoken to commercial devices. It also shows that the proposed method reduces the total time for interviewing subjects by 36% compared with the conventional method. We conclude that the proposed method can be used to build a useful corpus while reducing lead time and financial cost.

JANUS I I i s a r e s e a r c h s y s t e m t o d e s i g n a n d t e s t c o m pon e n t s o f s p e e c h t o s p e e c h t r a n s l a t i o n s y s t e m s a s w e l la s a r e s e a r c h p r o t o t y p e f o r s u c h a s y s t e m . W e w i l l f o c u s o n t w o a s p e c t s o f t h e s y s t e m : 1 ) n e w f e a t u r e s a n d r e f o r m a n c e o ft h e s p e e c h a n d 2 )t

For voice-enabled car navigation systems that use a multipurpose cloud speech recognition service (cloud ASR), utterance classification that is robust against speech recognition errors is needed to realize a user-friendly voice interface. The purpose of this study is to improve the accuracy of utterance classification for voice-enabled car navigation systems when inputs to a classifier are error-prone speech recognition results obtained from a cloud ASR. The role of utterance classification is to predict which car navigation function a user wants to execute from a spontaneous utterance. A cloud ASR causes speech recognition errors due to the noises that occur when traveling in a car, and the errors degrade the accuracy of utterance classification. There are many methods for reducing the number of speech recognition errors by modifying the inside of a speech recognizer. However, application developers cannot apply these methods to cloud ASRs because they cannot customize the ASRs. In this paper, we propose a system for improving the accuracy of utterance classification by modifying both speech-signal inputs to a cloud ASR and recognizedsentence outputs from an ASR. First, our system performs speech enhancement on a user's utterance and then sends both enhanced and non-enhanced speech signals to a cloud ASR. Speech recognition results from both speech signals are merged to reduce the number of recognition errors. Second, to reduce that of utterance classification errors, we propose a data augmentation method, which we call "optimal doping," where not only accurate transcriptions but also error-prone recognized sentences are added to training data. An evaluation with real user utterances spoken to car navigation products showed that our system reduces the number of utterance classification errors by 54% from a baseline condition. Finally, we propose a semi-automatic upgrading approach for classifiers to benefit from the improved performance of cloud ASRs.