Wrote a new NetEq unit test to test a network freeze during comfort
noise playout. The network freezes and resumes during the silence
period, and then resumes speech. It was verified that the delay
increased due to the freeze, and this CL contains a fix for that
problem.
BUG=2995
R=turaj@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/9849004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5701 4adac7df-926f-26a2-2b94-8c16560cd09d
This CL implements a unit test to cover an case where comfort noise
packets should be discarded. The situation arises when NetEq gets a
duplicate comfort noise packet. Without this check, the duplicate would
be decoded, and a the timing would shift.
As it turned out, the corner-case funcionality was not completely
accurate in NetEq4. This is because decision_logic_::cng_state_ is set
after the corner-case check. In the old NetEq3, the corresponding state
was changed before the check. This is now fixed.
R=turaj@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/9639005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5685 4adac7df-926f-26a2-2b94-8c16560cd09d
std::memcpy -> memcpy for instance. This change was motivated by a
compile report complaining that std::rand() was used instead of rand(),
probably with a stdlib.h include instead of cstdlib. Use of C functions
without the std:: prefix is a lot more common, so removing std:: to
address this.
BUG=
R=tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/9549004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5658 4adac7df-926f-26a2-2b94-8c16560cd09d
The variable playout_timestamp_ was not updated to the latest decoded
timestamp while comfort noise was played. Instead, it was upadted using
dead reckoning, which caused it to drift away from the timestamps of the
incoming CNG packets. Now it is updated also during comfort noise
playout.
Since the change is only in NetEq4, this change also makes the test
PlaysOutAudioAndVideoInSync use both ACM1/NetEq3 and ACM2/NetEq4.
Re-enabling one NetEq unit test that is no longer failing thanks to this CL.
BUG=2932
R=stefan@webrtc.org, turaj@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/8799004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5649 4adac7df-926f-26a2-2b94-8c16560cd09d
Converting the test to a method within the test fixture, and setting
up two tests that call this method. One for positive and one for
negative clock drift. The one with positive clock drift is disabled
for now since it does not pass, but will be re-enabled shortly.
This change is only made for NetEq4.
R=tlegrand@google.com
Review URL: https://webrtc-codereview.appspot.com/8599004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5541 4adac7df-926f-26a2-2b94-8c16560cd09d
We encounter a sample-underrun if NetEq is initialized with a sampling rate fs =16000 and receive Opus packets with frame-size less than 5 ms. The reason is as follows.
Let say NetEq buffer has 4 packets of Opus each of size 2.5ms this means that internally timestamp of packets incremented by 80 (internally Opus treated as 32 kHz codec). Given the initial sampling rate of NetEq, at the first time that it wants to fetch packets, it targets to fetch 160 samples. Therefore, it will only extracts 2 packets. Decoding these packets give us exactly 160 samples (5 ms at 32 kHz), however, upon decoding the first packet the internal sampling rate will be updated to 32 kHz. So it is expected that sync buffer to deliver 320 samples while it does only have 160 samples (or maybe few more as it starts with some zeros). And we encounter and under-run.
Even if we ignore the under-run "assert(sync_buffer_->FutureLength() >= expand_->overlap_length())" (neteq_impl.cc::811) is trigered. I'm not sure what happens if we remove this assert perhaps NetEq will work fine in subsequent calls. However the first under-run is blocking ACM2 test to pass.
Here I have a solution to update sample rate as soon as a packet is inserted, if required. It not a very efficient approach as we do the same reset in NetEqImpl::Decode().
It is a bit tricky to reproduce this because the TOT ACM tests do not run ACM2. In https://webrtc-codereview.appspot.com/2192005/ I have a patch to run both ACMs. To reproduce the problem, one can patch that CL and run
$ out/Debug/modules_tests --gtest_filter=AudioCodingModuleTest.TestOpus
Note that we would not encounter any problem if NetEq4 is initiated with 32000 Hz sampling rate. You can test this by setting |kNeteqInitSampleRateHz| to 32000 in webrtc/modules/audio_coding/main/acm2/acm_receiver.cc
BUG=
R=andrew@webrtc.org, henrik.lundin@webrtc.org, kjellander@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2306004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4896 4adac7df-926f-26a2-2b94-8c16560cd09d
Having this failing test being disabled in code will make it
possible to add it on the bots again, and make thus no bot
configuration update needs to be communicated when it's fixed.
BUG=1458
TEST=Compiled with GYP_DEFINES=target_arch=x64 and ran the
test successsfully on Windows. Also ran regular trybots.
R=tina.legrand@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/1593004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4163 4adac7df-926f-26a2-2b94-8c16560cd09d
However, two other "hacks" had to be added to maintain bit-exactness
with legacy.
Note that this change requires a new version of the universal.rtp test
input, although the output reference stays the same.
Moving reference files, and using a new input vector for NetEq4.
The new input vector neteq_universal_new.rtp is identical to the old
neteq_universal.rtp, except that the payload type for CNG packets that
follows a wideband codec is changed to 98.
Update to resources revision 15 where the new reference files are.
Also changing a faulty log error.
Review URL: https://webrtc-codereview.appspot.com/1078009
git-svn-id: http://webrtc.googlecode.com/svn/trunk@3442 4adac7df-926f-26a2-2b94-8c16560cd09d
