[Live-devel] MRCP over RTSP with "liveMedia" library

[Andriy Bychay]

Hello,

I'd like to use "liveMedia" library to implement MRCP over RTSP. I
read the FAQ, looked through the samples and traced library sources
and I still have some questions:

1. client establishes session as usual and streams audio data to
speech recognition server (unicast PCMU audio streaming).
For now I didn't found how to set data source and sink with RTSPClient
class to send data. If I  do something similar to client samples and
call MediaSubsession::initialize() method then it creates source to accept
data - the opposite thing.

Can you give me a hint?

2. client sends audio data on their arrival (for example, data are
acquired from microphone in real time); should I start
TaskScheduler::doEventLoop() every time I need to send acquired audio
and exit this loop if my data are over or there is better way to do
this?
Possibly I sould never exit the loop while I need to stream data from
the point of "liveMedia" library view. But between sending data I need
to send some commands to recognition server with ANNOUNCE messages.

3. As soon as server generates some decision on the data that client sent
it should send a response in body of ANNOUNCE RTSP message. Can client
accept such a message? How to do this?

-- 
Best regards,
 Andriy                          mailto:andriy.bychay at clarity-ag.net