RE: How do they record the actors for voice response systems

Subject: RE: How do they record the actors for voice response systems
From: Stuart Burnfield <slb -at- westnet -dot- com -dot- au>
To: Techwr-l <techwr-l -at- lists -dot- techwr-l -dot- com>
Date: Tue, 7 Apr 2009 13:18:00 +0800 (WST)

It would just be a case of recording and storing a separate set of numbers for each intonation, and having the software select the voice file to be played depending on the position of the number.

So for an initial prototype imagine that you record the numbers just once and store them in a set of files called,, The software is developed to extract numbers from the database and play them back in order using the voice files.

So far, so good, but it sounds a bit flat.

You record a second set of files with the intonation you would expect if the number were the last digit. You call these voice files,, etc, or perhaps use, etc, same as before, but store them in a folder called 'last'. Now the developers adjust the software so that when it gets to the last digit it uses the voice file in the 'last' folder.

You can get as fancy as you like--record a set of first digits, program in pauses at logical break points, record pairs of digits ("nine-two, four-three, six-seven, two-one"), and so on.

I worked on a trial of an early IVR system for the Australian Bureau of Statistics in about 1990. It didn't go anywhere (nothing to do with the technology) but it was a lot of fun.


> I was thinking most of the messages from the phone company
> that recite phone numbers. What has impressed me is that they
> approximate natural intonation to the point of having the
> intonation be *different* for the first and last digit of the
> phone number. Connie was able to use a *constant* tone of voice
> so that the end result sounded reasonable, but the messages I'm
> thinking of actually go so far as to have intonations that *vary*
> in line with natural intonation -- e.g., to go down on the last
> digit and "straight across" on the others. I was wondering if
> anybody knew specifically how that was accomplished.

ComponentOne Doc-To-Help 2009 is your all-in-one authoring and publishing
solution. Author in Doc-To-Help's XML-based editor, Microsoft Word or
HTML and publish to the Web, Help systems or printed manuals.

Help & Manual 5: The complete help authoring tool for individual
authors and teams. Professional power, intuitive interface. Write
once, publish to 8 formats. Multi-user authoring and version control!

You are currently subscribed to TECHWR-L as archive -at- web -dot- techwr-l -dot- com -dot-

To unsubscribe send a blank email to
techwr-l-unsubscribe -at- lists -dot- techwr-l -dot- com
or visit

To subscribe, send a blank email to techwr-l-join -at- lists -dot- techwr-l -dot- com

Send administrative questions to admin -at- techwr-l -dot- com -dot- Visit for more resources and info.

Please move off-topic discussions to the Chat list, at:

Previous by Author: RE: Question for Flare Users
Next by Author: RE: Training Manuals
Previous by Thread: Re: How do they record the actors for voice response systems
Next by Thread: what's this called

What this post helpful? Share it with friends and colleagues:

Sponsored Ads