If one thinks about it, one is liable to get a headache. The ability of computers to talk to and understand responses from callers seems a little too futuristic to be believable, but it is a reality today that has made the efficiency of call centres and the services they provide higher than could have been imagined 20 years ago. The fact is that automated telephone systems have advanced far beyond a simple recorded message, so much so that, in some cases, natural conversations can effectively take place.
The technology used is very sophisticated and very intricate, with the precision of voice xml used to provide an efficient interactive experience for callers. Of course, a core component in such services is the speech recognition software that is utilised by centres, with its ability to turn the spoken word into text a factor that allows the computer programming to work so effectively.
Obviously, it is a computer system that manages the response units in a centre and directs the conversations between callers and the computer voice. There can be a bank of thousands of possible dialogues based on the possible responses that a caller might have to a prompt. Therefore, these computers need to have clear and concise programmes with which to run the systems.
VXML the chosen format for these interactive voice dialogues. The voice browser that is part of automated systems is designed to read VXML, initiating the exchange with the caller, understanding responses, providing options and distributing calls towards the correct destination.
This format is the standard used by all automated systems, with a vast range of services covered by it. Information can be attained, from flight tracking and order inquiries to driving directions and audio news magazines, while the precise nature of the programme instructions means that functions can switch from speech synthesis to dialog management to audio playback and a host of others.
It is a highly technical area, but in essence the VXML format allows for clear programming for a call centre computer to follow, enhancing the services provided and ensuring the minimum possible confusion between technology and human when interaction actually begins. The quality of the computerised voice is remarkable too, with the stereotypical monotone robot voice now a dated feature. Modern systems create a more natural tone in order to achieve a more personable service for callers.
Technology that can recognise the spoken word first came to public attention in the mid 1960s, but it has only been in the last couple of decades, with the advancement of interactive technologies, that has seen it applied in any truly practical way. The fact that call centres must deal with the spoken word all day means that this technology is perfectly suited to the sector. But while earlier versions could log the voice pattern of one voice, as used in security systems, the technology used in call centres is capable of recognising arbitrary voices.
What this means is that random voices can be understood, not just the voice of a specified person, and with the wide range of voice tones, accents and pronunciations even within the English language, that is a major development. So, regardless of whether a caller is from Edinburgh, London, Dublin, Sydney or New York, the system can understand the responses that the caller may give when prompted by the interactive system. The technology is used in voice dialing, call routing, domotic appliance control and in processing speech to text, amongst several other functions.
However, when used in unison with an automated telephone service, the result is a near seamless, conservation between computer and caller. The words spoken by the caller are translated into text, which is then matched with the specific voicexml formatted response. That response is then played by the computer through a humanised voice, and the subsequent caller response is similarly translated.
Of course, this system is only applicable where a self service ability exists, such as when a simple option needs to be selected or a list of transactions to be detailed. While speech recognition software can be used to initially determine whether a caller needs to speak to a call centre agent or not, it is not necessary when an agent is speaking directly to the caller.