The sample uses are from when I debugged bug 617124. The change in neteq_network_stats_unittest.cc is a fix for a minor unrelated bug found by the try bots when I tried to land this CL (a test was passing uninitialized packet data to NetEq).
BUG=chromium:617124
Review-Url: https://codereview.webrtc.org/2293893002
Cr-Commit-Position: refs/heads/master@{#14034}
If a CNG packet is received first, followed by a speech packet with
another sample rate, NetEq should treat this as a change of codec, flush
out the CNG packet and reset the sample rate to that of the speech
packet.
BUG=webrtc:5447
NOTRY=True
Review-Url: https://codereview.webrtc.org/2307493002
Cr-Commit-Position: refs/heads/master@{#14032}
That is, rather than keeping a separate pointer and size.
This helps automate memory management in NetEq and will be useful in the
work to minimize the AudioDecoder interface as part of the injectable
audio codec work.
I'm planning a follow-up that will change the current management of Packet* to wrapping them in unique_ptr instead.
Review-Url: https://codereview.webrtc.org/2289093003
Cr-Commit-Position: refs/heads/master@{#14002}
The compiler optimization for Windows is O1 by default in GN, but O2 in GYP.
This might help explain the regression observed on neteq_performance_unittest.
NOTRY=True
BUG=641966
Review-Url: https://codereview.webrtc.org/2291253003
Cr-Commit-Position: refs/heads/master@{#13999}
With this change, the value 0xFF is no longer used to flag that the RTP
type is unknown. Instead, an empty value for the rtc::Optional is used.
Review-Url: https://codereview.webrtc.org/2290153002
Cr-Commit-Position: refs/heads/master@{#13989}
In order to get resource files to be properly packaged into
the .app for a unit test on iOS, the resource files needs
to be listed as sources in a bundle_data target.
BUG=webrtc:5949
NOTRY=True
Review-Url: https://codereview.webrtc.org/2292853002
Cr-Commit-Position: refs/heads/master@{#13968}
This adds a new file, webrtc/modules/audio_coding/neteq/tools/packet_source.cc, so that I'll have somewhere to put the new non-inlined methods.
NOTRY=true
BUG=webrtc:163
Review-Url: https://codereview.webrtc.org/2290593002
Cr-Commit-Position: refs/heads/master@{#13956}
This is still a tiny lie, since it checks for kCodecArbitrary to avoid
scaling a codec with an external decoder, because of missing information
in that case. The main point is still true, though. Once the next CL is
in, removing NetEqDecoder usage completely in DecoderDatabase, another
workaround will be in place for external decoders until we can fully
deprecate them.
BUG=webrtc:5805
Review-Url: https://codereview.webrtc.org/2270063006
Cr-Commit-Position: refs/heads/master@{#13952}
the number of points that need to be mocked for testing.
For the now non-virtual methods, DecoderDatabase now does a lookup
through GetDecoderInfo and then delegates to the appropriate method in
the DecoderInfo object, if one is found.
A few other methods were also changed to look up through GetDecoderInfo.
Also moved the audio decoder factory into DecoderInfo, so that
DecoderInfo::GetDecoder can be used directly.
Review-Url: https://codereview.webrtc.org/2276913002
Cr-Commit-Position: refs/heads/master@{#13933}
We have RTC_CHECK and RTC_DCHECK for C now, so we should use it. It's
one fewer difference between our C and C++ code.
NOPRESUBMIT=true
Review-Url: https://codereview.webrtc.org/2274083002
Cr-Commit-Position: refs/heads/master@{#13930}
Checksums were updated for NetEq and ACM bitexactness tests (after
verifying the audio quality).
BUG=webrtc:5447
Review-Url: https://codereview.webrtc.org/2266293005
Cr-Commit-Position: refs/heads/master@{#13928}
This removes the warning printouts about unknown header extensions.
BUG=webrtc:2692
Review-Url: https://codereview.webrtc.org/2266403005
Cr-Commit-Position: refs/heads/master@{#13912}
The current_rtp_payload_type_ should only be updated when the packet is
actually inserted into the packet buffer, since then the payload type
has been validated. This CL removes an unvalidated setting of this value
that happened after SSRC change or upon first packet.
BUG=webrtc:5447
Review-Url: https://codereview.webrtc.org/2270793003
Cr-Commit-Position: refs/heads/master@{#13910}
If an error happens in the GetAudio call, for instance when corrupt
payloads are inserted, GetAudio wil return an error. In this case, the
audio frame is not always correctly populated, which is to be expected.
BUG=webrtc:5447
Review-Url: https://codereview.webrtc.org/2272963002
Cr-Commit-Position: refs/heads/master@{#13902}
This implementation interprets payloads of size 1 as codec-internal SID
frames, marking the start of a CNG period. Changes were made to other
parts of the test payload chain, since it had to make use of the virtual
payload size in the case of header-only RTP files.
BUG=webrtc:2692
Review-Url: https://codereview.webrtc.org/2275903002
Cr-Commit-Position: refs/heads/master@{#13901}
iOS tests packaged into an .app uses the same way of
defining resources (the data attribute). Some iOS
simulator tests are failing due to missing resources, so
let's sync them all.
BUG=webrtc:5949
NOTRY=True
Review-Url: https://codereview.webrtc.org/2277753003
Cr-Commit-Position: refs/heads/master@{#13898}
We detect an unreasonable state (caused by a bad encoded stream)
before it can lead to problems, and handle it by resetting the
decoder.
NOPRESUBMIT=true
BUG=chromium:617124
Review-Url: https://codereview.webrtc.org/2255203002
Cr-Commit-Position: refs/heads/master@{#13888}
So that we don't have to use assert(). Includes one sample call site.
NOTRY=true
BUG=chromium:617124
Review-Url: https://codereview.webrtc.org/2262173002
Cr-Commit-Position: refs/heads/master@{#13862}
The new method returns the current total delay (packet buffer and sync
buffer) in ms, with smoothing applied to even out short-time
fluctuations due to jitter. The packet buffer part of the delay is not
updated during DTX/CNG periods.
This CL also pipes the new metric through ACM and uses it in
VoiceEngine. It replaces the previous method of estimating the buffer
delay (where an inserted packet's RTP timestamp was compared with the
last played timestamp from NetEq). The new method works better under
periods of DTX/CNG.
Review-Url: https://codereview.webrtc.org/2262203002
Cr-Commit-Position: refs/heads/master@{#13855}
scale1 == 31 if and only if w10 == 0. So even though 1 << scale1
overflows, we know that the result of the multiplication should be 0.
Handle that case.
BUG=chromium:615818
Review-Url: https://codereview.webrtc.org/2258543002
Cr-Commit-Position: refs/heads/master@{#13847}
Take 'tools/neteq_quality_test.cc' and 'tools/neteq_quality_test.h' outside of neteq_test_support into their own target, neteq_quality_test_support.
BUG=webrtc:6228
NOTRY=True
Review-Url: https://codereview.webrtc.org/2252413002
Cr-Commit-Position: refs/heads/master@{#13831}
When the sanitizer bots are switched to GN, this needs to be included as a dependency so that the executables can be compiled.
BUG=webrtc:6215
NOTRY=True
Review-Url: https://codereview.webrtc.org/2250893003
Cr-Commit-Position: refs/heads/master@{#13829}
The cast involves packet_len_samp, which is derived from the timestamps
and sequence numbers of incoming packets. Being values from the outside,
these should be treated as if any value is possible, making a
checked_cast unsuitable. Changing instead to saturated_cast to avoid
overflow with out-of-bounds values.
Review-Url: https://codereview.webrtc.org/2243403007
Cr-Commit-Position: refs/heads/master@{#13815}
Reland of https://codereview.webrtc.org/2072753002/ which broke
chromium due to how their build was setup. This issue should now be
resolved.
Changed WebRtcVoiceEngine to present receive codecs from the formats
provided by its decoder factory. Added supported formats to
BuiltinAudioDecoderFactory. Added helper functions for creating some
simple decoder factories for mocking.
Created a PayloadTypeMapper for assigning payload types to formats. I
think we'll eventually want to use this further up, or possibly in
both the audio and video sides. It would be best if the engines didn't
have to talk payload types at all, but it might be more difficult to
get around when payload types depend on each-other, like the RTX
codecs for video.
BUG=webrtc:5805
TBR=ivoc@webrtc.org
Review-Url: https://codereview.webrtc.org/2250683002
Cr-Commit-Position: refs/heads/master@{#13793}
