WebRTC[24]-WebRTC M75 Release Notes
目录
前言
WebRTC M75同样提供了Android和iOS的版本库，它包含3个新功能和50多个错误修复，增强了相关的功能和稳定性，以及一些性能的改进措施。
重要特性
1. RTPsenders和RTPReceiver支持了IEC和DTLS的设置接口。
2. chrome://webrtc-internals包含了标准的getStats()信息以及其他一些新的状态信息。
3. 完善了PeerConnection的negotiationneeded事件通知机制，之前的版本通知应用程序需要进行会话协商存在一定问题。
修复内容
| Issue |
Description |
Component |
| Simulcast streams will send one key frame on all spatial layers for each FIR with different SSRC |
Video |
|
| Makes send packet information non optional for feedback reports. |
BWE |
|
| Unify congestion window and pacing buffer pushbacks. |
BWE |
|
| Include pacing buffer size in congestion window pushback. |
BWE |
|
| Create FrameBufferController interface and allow its injection for VP8 |
Video |
|
| Make AEC3 the default AEC option in WebRTC |
Audio |
|
| H264 constrained baseline fails to be decoded |
Video |
|
| Excessive AEC suppression |
Audio |
|
| Pass coded_size info along in webrtc::VideoFrame |
Video |
|
| Bandwidth toggles between two estimates in StartUpPhase. |
BWE |
|
| PacedSender send to much padding when there are small packets sent |
BWE |
|
| AEC3: Echo during onsets |
Audio |
|
| RTT based backoff is not capped below. |
BWE |
|
| Simulcast video sends SDES with CNAME items with zero length |
Network>RTP, PeerConnection |
|
| Postpone decoding after expand causes too much delay in high packet loss scenarios |
Audio |
|
| Add support for writing a call order file in audioproc_f |
Audio |
|
| Fuzzing for simulcast |
PeerConnection |
|
| Add histograms to bandwidth probing code |
BWE |
|
| Increase default maximum jitter buffer size |
Audio |
|
| Make keyframe generation/request intervals tuneable |
Video |
|
| The way OpenSLEngineManager is shared between OpenSLESPlayer and OpenSLESRecorder is unsafe |
Audio |
|
| Adopt INTER_LAYER_PRED_OFF |
Video |
|
| Provide common interface for bitstream parsers |
Video |
|
| [standard stats] Implement counters for retransmitted bytes |
Stats |
|
| [standard stats] Implement totalEncodeTime |
Stats |
|
| [standard stats] Implement lastPacketReceivedTimestamp |
Stats |
|
| [standard stats] Implement stat for content type |
Stats |
|
| Acknowledged bitrate estimate can get stuck at low bandwidth. |
BWE |
|
| AEC3: missing bound checks when accessing a vector in the signal dependent erle estimator code |
Audio |
|
| RTCP XR target bitrate could be incorrect. |
Video |
|
| addTransceiver doesn't validate input rids |
PeerConnection |
|
| Fix timeouts in replay fuzzers. |
Blink>WebRTC>Network |
|
| Pass information from incoming LossNotification RTCP messages to video encoder |
Video |
|
| Duration of video pause is not included into sum of squared frame durations |
Video |
|
| The minimum comfort noise level in AEC3 is too high |
Audio |
|
| Color space not parsed correctly on receiver side |
Network>RTP |
|
| The runtime-settings in aecdumps for the pre-amplifier gain cannot be overruled in audioproc_f |
Audio |
|
| AEC3: Linear output used in suppression gain computation in non-linear mode |
Audio |
|
| Incoming offer for simulcast does not generate video |
PeerConnection |
|
| Duplicate calls to OnSentPacket() breaks ALR detection |
BWE |
|
| In simulcast mode VP9 sender doesn't write scalability structure on key frames of high spatial layers |
Video |
|
剩余60%内容，订阅专栏后可继续查看/也可单篇购买
WebRTC 作为当下最热门的实时音视频通讯框架，涉及非常多的过程，比如采集、编码、组包、发包、传输、收包、丢包重传、解封装、解码、音视频同步、渲染等，同时还包括很多功能特性，比如ANS、AGC、AEC，REMB、GCC、CNG、FEC、PLI、SVC等，需要一点点深入理解其中的奥秘。
