subject: Can Voice Recognition Automate Transcription Service? [print this page] The companies that have built voice recognition software claim "Yes", but professional users are saying "No", they maintain that voice recognition is fine if it is limited to documenting a couple of minutes of audio. When large file size audio recordings have to be converted into text, the software falters and fails to deliver. The voice recognition has a long way to go, for it to be used to automate transcription service, according to professional users of the software.Many television companies which started using voice recognition for transcribing digital voice recordings of the program are disappointed, because of the low accuracy levels and the time it takes to transcribe the audio. These companies use the transcripts for adding captions to television programs, and for the captions to sync with the audio, the accuracy of the transcripts required is 100%. This system disappointingly fails on this account.The recorded data used by researchers from interviews and focus groups also have experimented with speech recognition and some of them appeared to be satisfied with the performance of the software. They maintain that the application has to be trained for speech recognition and it is very convenient for transcribing short notes. However for large size recordings they still depend on transcription companies to document recorded data. There are inherent deficiencies in the application which have to be figured out before large scale use of the application becomes possible.An attorney working for a legal company tried the voice recognition for transcribing recordings of court-room proceedings, but the end results were disastrous according to him. The voice recognition failed to transcribe a couple of minutes of the proceedings, legal terms were mispronounced and finally he had to call his regular transcription company for documenting the proceedings. The attorney did not want to use the application at all, even for the transcription of short notes.The commonality of views expressed by different professionals does indicate that commercial use of voice recognition software in not possible, there are limitations of what it can do and what it cannot. The transcription services for the moment shall remain in the human domain, and automation will have to wait until the technology is mastered for commercial use.
