Voice-recognition technology has long been heralded as an ultra-futuristic luxury. Even as early as the 1960s, the public recognized the potential of voice-recognition technology. The original ’60s Star Trek series saw the Enterprise crew navigating through space with a voice-activated computer that could also read back relevant data to help Captain Kirk more effectively command the vessel. Back then, it was deemed a figment of science fiction, but surprisingly, just 50 years later, that figment of science fiction is closer than ever to becoming a reality.
Voice-recognition technology has been largely hit or miss up to this point. Phones that are programmed to recognize voice command prompts, such as "call mom," often misunderstand the command dial someone else instead. Many automated phone systems will ask users on the other end to repeat pieces of information again and again because the technology cannot recognize that is being said. All in all, voice-recognition technology has undoubtedly caused many people to have screaming matches with their machinery due to its failure to accurately recognize different voice prompts. Program developers have been working tirelessly to fix that problem and their efforts are beginning to show. One solution that has been developed is to lessen the number of voice commands that a phone can recognize. While this may seem like the opposite of what the technology should progress towards, it actually lessens the margin for miscommunication. Fewer command prompts means that instead of shouting out ten different versions of a command in the hopes that the device will recognize one of them, users can choose from a smaller library of commands. In addition, if all of these commands sound dramatically different, the chances are greater that the device will get the command right the first time around.
Other aspects of voice-recognition activity are improving as well. Transcription technology, such as the types of programs used by legal clerks when they transcribe court proceedings, is being improved by providing the program with a database of frequent word combinations and usages. Whenever someone speaks a sentence into the microphone, the program will compare each word to the database to see the most common ways in which those words are used so that the transcribed sentence will be as accurate as possible. Lightning-fast computers are also making transcription programs function more efficiently.
However, even as the technology gets better and better, we are still a long way from having a "conversational computer," which means that it will still be many years before voice technology can reach the levels of the Enterprise’s talking computer.
Leave Comments