Back to top
- What file format should I use for audio files?
u-law encoded WAV files, 8,000 kHz, 8 bits per sample, mono.
You can record prompts using Windows Sound Recorder. Save the files as described above, selecting ‘CCIT u-law’ as the Format.
If you’re running Vista, Sound Recorder will only let you record WMA files. You can use Audacity instead.
Audacity is also great for editing recorded prompts.
Audacity or not, be sure to use the ‘Find zero crossings’ feature when trimming audio clips – otherwise you will end up with audible ‘clicks’ at the beginning and/or end of the clip.
- What is a CallFlow?
It’s a bit of XML you can use to interact with the CallFlow service. You can pass it directly to the service to trigger an outbound call, provision it so that it is available when an inbound call is received, or generate it dynamically at runtime in response to, say, user input such as touchtone digits.
Here is a very simple callflow that places a call, plays an audio message to the callee, prompts them for touchtone input, and connects them to a party if they press 1:
<?xml version="1.0" encoding="UTF-8"?>
<callflow
xmlns="http://sdk.bt.com/callflow/2007/04">
<call id="start" target="tel:xxxxxxxxx" next="hi"/>
<announcement
id="hi"
audio="welcome.wav"
next="select_option"/>
<prompt id="select_option" audio="connect_prompt.wav">
<input pattern="1" next="connect"/>
<default next="select_option"/>
</prompt>
<dialog id="connect" target="tel:xxxxxxxxx"/>
</callflow>
- Do we really need another XML language for voice? What’s wrong with VoiceXML / CCXML?
Web21C SDK and services have been built by developers, for developers. Here is our list of VoiceXML / CCXML dislikes, which we believe CallFlow addresses:
- There are two of them!
- They have not truly delivered the promise of interoperability
- The learning curve – whereas a developer with no background in telecommunications can get up to speed in hours, VoiceXML and CCXML this is measured in months
- Mashup-unfriendliness
- Security – with VXML / CCXML, this is left to the transport. We believe we can provide a simpler solution, and will shortly be adding to our feature set shortly.
- Programming model – VXML’s design reflects the fact that it was designed to provide ‘voice browsing’ functionality. We believe that voice is a poor medium for browsing information, analogous to viewing a webpage through a keyhole. We further believe that offloading this baggage, leaving browsing to the browser, and focusing on making the creation of voice-based interactions both simple and easy to integrate with applications leads to a representation that developers prefer
|
