ABSTRACT: Automatic speech recognition is not perfect. There will always be a certain amount of recognition errors involved; be it because of individual dialects, speaking style, seasonal buzz words or background noise. To ensure an accurate transcript,
manual correction is necessary. With Speechless, correction times are as low as 4 times of the media length, while traditional transcription at Sail Labs Technology takes a trained user about 10 times as long. To provide for a continuous improvement of the transcribing system, a concept of classifying and correcting errors in a transcript was needed. The desired software should be designed as an option for manual classification and correction of errors in the transcript. The general requirements were a faster correction speed, the possibility of classifying errors in the transcript and an interface which is easy to use.
To access the complete research paper, please fill in the form below.
Your name and e-mail are going to be used in order to send you only the research file and not any additional commercial material. You can change your mind at any time by clicking the unsubscribe in the footer of the email that you receive from us, or by contacting dataprotectionofficer@hensoldt-analytics.com. Please find out about your rights and choices and how we use your information in our Privacy Policy.
