News – Speech-recognition Technologies

The latest from the BBC

The BBC today reports on developments in speech-recognition technologies. While these have advanced over the years, the voice-recognition software found in computers, handsets and sat-nav systems, among others, is far from perfect. Full speech recognition is incredibly complex because it requires the understanding of any word said in any sequence, and the scale of that challenge perhaps explains why technology still struggles to understand language comprehensively.

Software remains sluggish when compared with the brain. When words are missed or slurred, the brain can use context to understand what is meant. Steps towards enabling computers to do the same are being made as computers gain processing power and memory. Modern speech-recognition systems do now employ statistical modelling and context to work out what is being said, using identifiable words to disambiguate those that are less clear.

Computers interpret speech by breaking down words into small chunks of sounds called phonemes, the building blocks which make up English and every other language. The latest speech-recognition programs may no longer need the user to train them to understand particular speech patterns or voices, but that is not the end of the problems that speech presents. Environmental factors, such as background noise, can be detrimental to the technology’s accuracy. To overcome this, programs still require human oversight to decipher meaning.

Another problem facing computerised systems is unfamiliar words, the result of a language’s natural evolution as words acquire new meanings. For the experts cited in the BBC report, new terms like ‘staycation’ emphasise the importance of human knowledge in understanding speech. As the BBC concludes, ‘For the moment, a machine that fully comprehends what people are saying remains in the realms of science fiction.’

Global Lingo, a leading transcription company, uses only highly trained professional writers to guarantee the accuracy and fluency of its transcripts. For more information, please call one of the Global Lingo team today on +44 (0)20 7870 7100.