Use Of Automatic Speech Recognition Research Articles

Errors and error correction in automatic speech recognition systems

Abstract Errors, whether created by the user, the recognizer, or inadequate systems design, are an important consideration in the more widespread and successful use of automatic speech recognition (ASR). An experiment is described in which recognition errors are studied under different types of feedback. Subjects entered data verbally to a microcomputer according to four experimental conditions: namely, orthogonal combinations of spoken and visual feedback presented concurrently or terminally after six items. Although no significant differences in terms of error rates or speed of data entry were shown across the conditions, analysis of the time penalty for error correction indicated that as a general rule, there is a small timing advantage for terminal feedback, when the error rate is low. It was found that subjects do not monitor visual feedback with the same degree of accuracy as spoken, as a larger number of incorrect data entry strings was being confirmed as correct. Further evidence for the use of ‘second best’ recognition data is given, since correct recognition on re-entry could be increased from 83·0% to 92·4% when the first choice recognition was deleted from the second attempt. Finally, the implications for error correction protocols in system design are discussed.

Feedback requirements for automatic speech recognition in the process control room

Automatic Speech Recognition (ASR) has great potential for use in control room systems; to date, there has been little research into the human factors issues this raises. For example, careful consideration needs to be given to the provision of adequate feedback to the user. We concentrate on the two main types of visual feedback: textual and symbolic. Two studies reported here show that little difference exists between them in user performance on a task requiring spoken control of a process. However, the results demonstrate a significant reduction in learning time when textual and symbolic feedback are combined. We defined the correction of device misrecognitions as a verbal decision task, for which Study 1 shows that textual feedback is most appropriate. However, Study 2 shows that textual feedback is more likely to be misunderstood than either symbols or a combination of text and symbols. A combination of both text and symbols is proposed as the most efficient form of feedback for the use of ASR in control room systems.

Automatic speech recognition in practice

Abstract There is a growing interest in the commercial possibilities offered by automatic speech recognition (ASR) technology. Unfortunately the prospective user has little independent guidance with respect to the potential success of any proposed implementation. There do exist a few general human factors guidelines on the use of ASR, but most of the corpus of knowledge that forms part of the lore within the ASR community is based on the unpublished experiences of system.developers and users themselves. The present paper attempts to redress this balance; it is a summary of the experiences of users and system designers at 30 research and commercial sites in the UK and USA where ASR has been extensively used or tested. The application represented were classified as vehicle, office, industrial, and aids for disabled people. A number of important human factors issues were identified, and the relative success of the various applications are discussed.

Improving speaker consistency in an automatic speech recognition framework

Inconsistencies in speaking pose a major obstacle to the effective use of automatic speech recognition (ASR) systems. In the present study, we sought to improve speaker consistency in ASR by means of instructions and feedback. The instructions were specific suggestions for effective interaction with ASR devices. Feedback was given by a visual barometer-like display, representing the similarity between an utterance and previous utterances. In two experiments, naive speakers were tested on isolated words using speaker-dependent templates. In the first experiment, subjects who were given instructions demonstrated greater speaker consistency than those who did not receive instructions. In contrast, feedback had very little effect on performance. Instructions then were revised, based on an extensive analysis of the types of errors made by speakers, and were tested in a second experiment. Results of this experiment showed a further improvement in recognition performance; the new-instructions group had recognition accuracies that were 6% higher than the no-instructions group.