voice-push.com

Pushing VoiceXML to the masses!



Home

Blog

VoiceXML
VoiceXML 2.0 & 2.1
VoiceXML 3.0
State Control XML
ASR & TTS
VoiceXML Applications

Video & VoiceXML
Video Apps

VUI vs. GUI
Client vs. Server Apps

Voice User Interfaces
DTMF vs. ASR
Target Audience
Dialog States
Global Commands
Zeroing Out
Personality
NLU vs. Directed Dialogs
Prompts - Wording
Prompts - Snippets
Prompts - Randomising
Prompts - Recording
Grammar Design
Waiting
Error Handling

Project Phases
User Requirements
Technical Spex
VUI Specifications
Development
Going Live!

Links

Contact

Who are you designing the application for?

This is one of the most important questions that you have to answer. Are you designing an application for teenagers that you expect to call at least once a week? Or are you designing a customer service application for senior citizens who may call once a year? The answers to these questions influence design issues such as: the kind of personality the application should have, whether ASR or DTMF should be used for input and whether it makes sense to use customer profiling to modify the VUI.

So for the purposes of demonstration lets have two applications. One is a fan site for a popular pop band, which has ring-tones to download, interviews with the band and some snippets of the tracks on their upcoming new release. The other application is a customer service application for pensioners who want to avail of an offer to have their telephone book delivered for free. Taking these two extreme cases and consider that implications they have.

What's the age group of your callers?

To put it bluntly the younger generation are more technically savvy than there grandparents - and they are more open to new technology. They are also more willing to be tolerant of something going wrong the first time they try it. So you for the fan-site application you could certainly consider ASR - in particular with DTMF fall-back for the case that their is a lot of back-ground noise. For the senior citizen’s application it probably makes more sense to go with DTMF - on the basis that this is pretty idiot proof. With DTMF it’s clear that you’re dealing with a machine, which can sometimes be confusing with speech rec applications. However, ASR could be of interest, if it was felt that senior citizens have more difficulty pressing DTMF keys due to smaller keypads, poor eyesight or reduced dexterity. If this is coupled with the fact that the basic answer to the question should either be yes or no, then the recognition rates should also be fairly high.

How often will they be calling?

When you’re writing an application that is only going to be called once or twice, then you have to explain everything to the caller in detail. It is important that they understand what the application does and what is expected of them. This may mean longer prompts, but this won’t annoy the caller as they won’t be calling the application repeatedly. So for our telephone book application it is obviously that we should be as clear as possible.

However, applications that are called regularly are a little trickier. So if we’re expecting the teenagers to call the fan-site application on a regular basis we can’t afford to drive them up the walls by repeating long prompts explaining what’s happening all the time. This is where user profiling comes in handy. User profiling mean keeping a record of what the caller has done - in its most simple form, how often they have called the application to date. So the first time a caller uses the application they hear the long prompts, which explain everything in great detail. However, there is a second set of prompts which are shorter and do less explaining. These can be used after the caller has used the application more than once or twice, or whatever you decide is best. In effect these are the power user prompts, which combined with power user commands can speed up a callers interaction with an application.

VUI Personality

This one is fairly easy. In the case of the telephone book application, the VUI should be polite and helpful. So a competent-sounding female voice in their thirties would probably quite suitable. In contrast to that the fan-site can afford to be more relaxed - and younger sounding. Ideally it would actually use one of the members of the group to record the prompts! Then it could also use slang - though the grammars would need to be tuned to responses that also use slang. This could prove to be a little tricky - particularly if the slang is continuously changing. In theory, the fan-site can also be witty - though this can be a dangerous route if callers use the system on a regular basis. A joke begins to wear thin pretty quickly - so handle with care.



If you have any comments, ideas, issues, etc. about this topic why not try the voice-push forums






© voice-push.com