Voice transcription software has become more and more common in recent years. Also known as voice recognition software, these transcription programs convert audio materials into text transcripts for use in other applications or to provide added formats for more convenient access to the information. While great strides forward have been made in these software applications, the transcripts produced through purely computerized methods are still not accurate enough to be used for most business, medical or professional purposes. Differences in accents and pronunciations of certain words continue to present significant problems during the transcription process. Additionally, words that sound alike are often used interchangeably by these programs, especially less sophisticated versions that lack comprehensive grammar-checking functions.
History of voice transcription software.
The first voice recognition program was introduced in 1952 and was capable of identifying single digit numbers when spoken. The IBM Shoebox was somewhat more advanced; released in 1961, it incorporated the same capabilities as the original program along with the ability to recognize the sixteen words programmed into its vocabulary. These first voice recognition programs were primitive and extremely limited in their functionality; later programs incorporated many more words and numbers in their online vocabulary. However, context and grammar were beyond most of these programs until very recently; even today, the word error rates of commercially available voice transcription software are unacceptably high. As a result, the transcripts created by these programs must be proofread and edited manually, robbing the software of much of its promised utility.
While voice transcription software lacks the accuracy and reliability needed for medical, scientific and business applications, voice recognition programs are commonly used in automated telephone systems. Because these computerized menus accept only a few preset responses, the opportunities for error are significantly decreased. Accuracy is not as important in these applications because callers are automatically routed to a human operator if the computer system repeatedly fails to understand their responses.
Advantages of human transcription over voice transcription software
No computer program currently available can compete with the linguistic abilities of the average person. Professional transcriptionists can draw on a wealth of personal experience and contextual knowledge in producing their transcripts; computer programs must rely solely on the information available within their code and vocabulary and cannot learn from past mistakes or analogous situations. While it is possible that voice recognition software will one day perform on the same level as human transcriptionists, currently, professional transcription providers still offer the most accurate and cost-effective services for general use.