Question

I am thinking to build a Telephone based search engine. The concept is simple:

  1. User Dials the number.
  2. We record his input and convert the speech into text.
  3. Use Google API to search for the query.
  4. Fetch the top results and convert them into speech.
  5. Send output to the user.

I'm comfortable in coding the mechanism. But i don't know how to implement this on a telephone line. I will need a IVR which will guide the user and a back end application for processing. I can code the backend application.

Can you guys please tell me how can I implement my app over a telephone line. I did some research and come up with something called Asterisk and VoiceXML. Is it possible to do the task using any of these methods.

Was it helpful?

Solution

If the question was "Is it possible to do the task using any of these methods?" then the simple answer is yes.

VoiceXML lets you define an application executed on a VoicePlatform like Cisco CVP, Avaya AVP, Genesys GVP, ... The only "issue" which you will have is that you need one of those for it (and they tend to be relatively expensive).

If you had a speech recognizer and a speech synthesis engine then you may be able to have calls coming into an Astrisk platform and sending these via SIP into a SIP-capable server which builds the dialog in any programming language you like (e.g. Java). Here you'll need a speech recognizer and a speech synthesis engine to do the "conversions".

Even though my answer implies it is simple, there are many issues to overcome on the way, like: speech quality, recognition accuracy, error handling, etc.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top