Given a sip call between two persons using freeswitch as my telephony engine ,how to catch audio stream of each person separately and process it before it's sent to the other end. Thanks for your help in advance.
2
votes
You mean record each leg separately?
– Sasi Varunan
No I want to process it real-time ...thanks for your comment
– Hataki
Would you mind sharing what processing you think to do? Freeswitch does some audio processing like transcoding, volume mixing in conference. Anything similar?
– Sasi Varunan
Not at all..it's a project in my college , I shall turn speech into text then translate it and finally turn it back into speech that the other end can hear..basically a simple voice translator for voip calls
– Hataki
I already set freeswitch up with both phocketsphinix as my speech recognition engine and flite as tts engine and tested them and every thing works fine ...now I want to know if I can combine them to do my project. Again I appreciate your interest.
– Hataki
1 Answers
1
votes
The only possible way i can think of is, Set up two conference. Originate a call to A and connect to Conf A on answer. call B and connect to Conf B. Now if A speaks you can record the call and convert to text - translate and convert to audio and play it to Conference B. Vice versa.
ESL is a powerful module in Freeswitch where you can able to get all the events of freeswitch application and play with. In conference you get events when a member speaks, Joins, leaves, Mute and so on. Its an Idea but i've not tried it.
Its like http://www.iamili.com/ that you gonna try :)
