Hi, I am working on a speech recognition application. All of my users will dial into an IVR I will provide. This IVR is based on Asterisk. I have now two option on how I allow users to reach to this IVR running on Asterisk system:
1) Telecom company PSTN (PRI) line terminates on TDM card connected to server running Asterisk. In this case, I am bound to have Asterisk on the server where the PRI line comes. Meaning I can not host Asterisk on remote server (Do you find any other option here?)
2) PSTN (PRI) line terminates on an IP PBX which gives IP channel connection to the server running Asterisk. In this case, I can host Asterisk on remote server, eg: Amazon Web Services or Linode etc.
As I read before, more the encoding and decoding happens, the natural voice coming on the PRI channel attracts noise to it and by the time it reaches to Asterisk, it has noise added from multiple sources/activities. Also, as I read somewhere (don’t remember the book/blog) that when Voice is carried over IP (usually UDP and SIP on top of it), it can add its own issues like: garbled speech or lost syllables within a word or sentence due to delay/loss of packets and/or jitter.
Now what do you suggest should I go with option 1 or option 2 mentioned above? The reason to ask is to know from experienced users if Voice On IP reaching to Asterisk and Voice via TDM to Asterisk has any difference (in terms of changing natural speech content)?