How to get the audio stream during a real-time conversation?

[298029lkk]

I want to get the voice of the other party in a stream.
Is there any good advice?

[siniypin]

Hello,

you can do so by plugging your own sound port implementation to the conference bridge, more info here:

• https://trac.pjsip.org/repos/wiki/FAQ#audio-man (note sink only media ports);
• https://trac.pjsip.org/repos/wiki/media-flow

#77

P.S. If you end up developing one it would be really great if you could share it with everyone.

[298029lkk]

@siniypin Thank you for your reply. According to your suggestion, is it necessary to modify the source code of pjsip?

[siniypin]

No, you don't need to modify pjsip, but you certainly need to plug into pjsip audio pipeline using the above mentioned tricks.

[siniypin]

One hack I can think of that wouldn't require pjsip integration is to stream audio to a file while simultaneously reading from the very same file. Never tried it, not even sure it is going to work for you.

[298029lkk]

Thank you for your support.
I want to get the other party's audio in real time after the phone is dialed in, then intelligently recognize and get the answer, and then send it to the user. Instead of a person speaking at the end of the phone.
This is what I want to achieve now. The key point is that I can't handle the acquisition and sending of audio.

[siniypin]

Well, you've got these two options. You can route an audio stream to a file with pjsip4net using a DSL builder method

[298029lkk]

Ok, thank you very much. But I think the second way can't achieve my purpose. Because I want a real-time conversation and a real-time answer. Instead of processing the recording file after the conversation is completed.
So should I use the first way? Where should the first way handle?
Also, if I have an audio stream answer now, how do I send it to the other party?

[siniypin]

I don't get what you aim for. It all sounds like you want to establish a common bi-directional call. pjsip4net does exactly that out of the box.