The very earliest days of text-to-speech solutions weren’t very noble. The responses, while comprehensible, were often comically stilted, which didn’t lead to TTS being a technology that consumers felt confident about. TTS, of course, has come a long way. While many companies have found efficiencies using text-to-speech interactive voice response (IVR) solutions, many of these companies found that they have had to maintain them more assiduously than they would a system that used simple recorded prompts.
Contact center technology and services company West Interactive (News - Alert) blogged recently with advice on how to spot problem areas in advance in TTS IVR solutions.
Misspellings. A text-to-speech engine is only as good as the data it “speaks” from. As a result, misspellings can be a real problem, and West Interactive suggests using spell-check solutions and extra care by agents that they have entered data properly (particularly names and other proper nouns that won’t show up in spellcheck). But it’s not only by error that misspellings occur.
“Sometimes purposeful misspelling is done, usually in an attempt to change the way the TTS engine says a word. This often backfires and makes the TTS engine sound awkward. A better solution is to create a user dictionary that tells the TTS engine exactly how to sound for certain word,” writes West.
Abbreviations. A TTS solution will encounter a lot of abbreviations, particularly in addresses. Rather than trying to eliminate abbreviations from all your data, West says there is a better solution. Just as can be accomplished with misspellings, abbreviations can be put into a dictionary and translated to their full words so the engine can read back an address with ease.
Addresses, telephone numbers and dates. TTS solutions may not always be able to handle numbers with confidence. Standard types of data like addresses, phone numbers and account numbers are often passed to the TTS engine from a variable in an application and are often out of context. West suggests a fix:
“Using Speech Synthesis Markup Language or SSML (the W3C standard) is a good way to handle this. Using tag (News - Alert) elements such as <say-as/> can tell the engine what type of data you’re feeding it,” writes West.
With a little extra care, companies can reap the efficiencies and cost savings of TTS-based IVRS while overcoming simple challenges, making the most of the technology in the call center.
Want to learn more about the latest in communications and technology? Then be sure to attend ITEXPO Austin 2012, happening now in Austin, TX. Stay in touch with everything happening at ITEXPO (News - Alert). Follow us on Twitter.
Edited by Jamie Epstein