A lot of people ask me whether I am worried about the development of speech recognition systems, which admittedly are getting more advanced year on year. Simply put, these systems have the ability to hear human speech and transfer it in real time into digital text, which can then be used in the same way a standard transcript would be.
I have looked into these systems over the years, and so far I haven’t seen anything that would make me nervous. Whilst such systems can be useful, especially for very long periods of audio, where only a very rough transcript is needed, this is really all they are currently good for.
Global Lingo is in the market of high-end transcription and added value services such as minute taking and well written reports. We provide documents that have been professionally written (by a human being) and then proofread by a second set of equally qualified set of eyes. It is not possible to write this kind of document using speech or voice recognition software. Our clients pay for this kind of service as they don’t want to have to trawl through every document they receive looking for errors (and believe me there will be many from a speech recognition document).
Below are some problems with speech recognition:
- The nature of narrative dictation is highly interpretive and often requires judgement that may be provided by a real human but not yet by an automated system.
- Another limitation has been the extensive amount of time required by the user and/or system provider to train the software.
- Speech and voice recognition systems find it difficult to recognise who is speaking when, which lets face it is quite important!
- These systems find subject matter of a technical matter very difficult. For instance, doctors reports on new drugs coming on the market which even Google hasn’t even heard of, speech recognition systems won’t know what to write.
- The grammar, punctuation and spelling in documents produced by these systems is usually pretty bad.
This is why I am not worried so far!