need help understand the echo cancellation code

[billconan]

:oops:

Hello there,

I'm trying to understand the echo cancellation code. but I can't make sense of the following code:

 

					QMutexLocker l(&qmEcho);

				if (qlEchoFrames.isEmpty()) {
					iJitterSeq = 0;
					iMinBuffered = 1000;
				} else {
					// Compensate for drift between the microphone and the echo source
					iMinBuffered = qMin(iMinBuffered, qlEchoFrames.count());

					if ((iJitterSeq > 100) && (iMinBuffered > 1)) {
						iJitterSeq = 0;
						iMinBuffered = 1000;
						delete [] qlEchoFrames.takeFirst();
					}
					echo = qlEchoFrames.takeFirst();
				}

 

I understand that the echo frames and the mic frames can have some offset. we need to somehow realign them. but I just can't understand iJitterSeq and iMinBuffered and why when (iJitterSeq > 100) && (iMinBuffered > 1) the queue needs to be popped?

also what do 1000 and 100 mean here? why did we hard code it to 1000?

later in the code, iJitterSeq is updated by this line:

 

			iJitterSeq = qMin(iJitterSeq + 1,10000U);
		qlEchoFrames.append(outbuff);

 

this means iJitterSeq should be the size of qlEchoFrames for most of the time (when the size is less than 10000)

and this line updates iMinBuffered

 

			iMinBuffered = qMin(iMinBuffered, qlEchoFrames.count());

 

meaning iMinBuffered should be also the size of qlEchoFrames for most of the time (when count is less than 1000)

does this mean iMinBuffered == iJitterSeq? then why do we use two variables?

Thanks a lot!