Fixing A/V sync issues with RTP/RTCP sent from mediasoup

Question

A little background: I'm attempting to record a webrtc call being made through the mediasoup v2 SFU. I'm using mediasoup's room.createRtpStreamer() method to generate a stream which mirrors RTP/RTCP to ffmpeg. Two streamers are created for audio and video within ~30ms of each other and begin broadcasting. FFmpeg then spins up and starts accepting. Pretty sure RTCP is working since ffmpeg is always starting with a keyframe despite being started after the streamer begins broadcasting.

The problem is that I encounter audio/video desynchronization with seemingly random offsets. My current theory is that this offset is based on how old the last keyframe is that RTCP requests to start the stream. See below for ffmpeg configuration and output but my question is: what ffmpeg arguments can I use to adjust the video frame timestamps to match the audio frame timestamps or vice versa? I've messed around with -map 0:0,0:1 -map 0:1,0:1 but it doesn't seem to do what I'm looking for.

ffmpeg flags:

'-y',
'-loglevel',
'debug',
'-dump',
'-protocol_whitelist',
'file,crypto,udp,rtp,data',
'-analyzeduration',
'20M',
'-probesize',
'20M',
'-i',
`data:text/plain;base64,${sdp.toString('base64')}`,
'-fflags',
'+genpts',
'-vcodec',
'copy',
'-acodec',
'aac',
'-bsf:v',
'h264_mp4toannexb',
'-start_number',
'0',
'-hls_list_size',
'2147480000',
'-hls_wrap',
'0',
'-hls_time',
'10',

SDP used for input (template):

v=0
o=- 0 0 IN IP4 <%=ip %>
s=title
c=IN IP4 <%=ip %>
m=audio <%=audioPort %> RTP/AVPF <%=audioPayload %>
a=sendrecv
a=rtcp-mux
a=rtpmap:<%=audioPayload %> opus/48000/2
a=fmtp:<%=audioPayload %> minptime=10; useinbandfec=1
m=video <%=videoPort %> RTP/AVPF <%=videoPayload %>
a=sendrecv
a=rtcp-mux
a=rtpmap:<%=videoPayload %> H264/90000
a=rtcp-fb:<%=videoPayload %> ccm fir
a=rtcp-fb:<%=videoPayload %> nack
a=rtcp-fb:<%=videoPayload %> nack pli
a=rtcp-fb:<%=videoPayload %> goog-remb
a=rtcp-fb:<%=videoPayload %> transport-cc
a=fmtp:<%=videoPayload %> level-asymmetry-allowed=1;packetization-mode=1

ffmpeg output - garbled with some timestamps

1512775954585 - stderr: ffmpeg version 3.4 Copyright (c) 2000-2017 the FFmpeg developers
  built with Apple LLVM version 9.0.0 (clang-900.0.38)
  configuration: --prefix=/usr/local/Cellar/ffmpeg/3.4 --enable-shared --enable-pthreads --enable-version3 --enable-hardcoded-tables --enable-avresample --cc=clang --host-cflags= --host-ldflags= --enable-gpl --enable-ffplay --enable-libmp3lame --enable-libvpx --enable-libx264 --enable-libxvid --enable-opencl --enable-videotoolbox --disable-lzma
1512775954587 - stderr:   libavutil      55. 78.100 / 55. 78.100
  libavcodec     57.107.100 / 57.107.100
  libavformat    57. 83.100 / 57. 83.100
  libavdevice    57. 10.100 / 57. 10.100
1512775954587 - stderr:   libavfilter     6.107.100 /  6.107.100
  libavresample   3.  7.  0 /  3.  7.  0
  libswscale      4.  8.100 /  4.  8.100
  libswresample   2.  9.100 /  2.  9.100
  libpostproc    54.  7.100 / 54.  7.100
Splitting the commandline.
Reading option '-y' ... matched as option 'y' (overwrite output files) with argument '1'.
Reading option '-loglevel' ... matched as option 'loglevel' (set logging level) with argument 'debug'.
Reading option '-dump' ... matched as option 'dump' (dump each input packet) with argument '1'.
Reading option '-protocol_whitelist' ...1512775954589 - stderr:  matched as AVOption 'protocol_whitelist' with argument 'file,crypto,udp,rtp,data'.
Reading option '-analyzeduration' ...1512775954590 - stderr:  matched as AVOption 'analyzeduration' with argument '20M'.
Reading option '-probesize' ...1512775954590 - stderr:  matched as AVOption 'probesize' with argument '20M'.
Reading option '-i' ... matched as input url with argument 'data:text/plain;base64,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'.
Reading option '-fflags' ...1512775954591 - stderr:  matched as AVOption 'fflags' with argument '+genpts'.
Reading option '-vcodec' ... matched as option 'vcodec' (force video codec ('copy' to copy stream)) with argument 'copy'.
Reading option '-acodec' ... matched as option 'acodec' (force audio codec ('copy' to copy stream)) with argument 'aac'.
Reading option '-vsync' ... matched as option 'vsync' (video sync method) with argument '0'.
Reading option '-map' ... matched as option 'map' (set input stream mapping) with argument '0:0,0:1'.
Reading option '-map' ... matched as option 'map' (set input stream mapping) with argument '0:1,0:1'.
Reading option '-bsf:v' ... matched as option 'bsf' (A comma-separated list of bitstream filters) with argument 'h264_mp4toannexb'.
Reading option '-start_number' ...1512775954591 - stderr:  matched as AVOption 'start_number' with argument '0'.
Reading option '-hls_list_size' ...1512775954591 - stderr:  matched as AVOption 'hls_list_size' with argument '2147480000'.
Reading option '-hls_wrap' ... matched as AVOption 'hls_wrap' with argument '0'.
Reading option '-hls_time' ...1512775954591 - stderr:  matched as AVOption 'hls_time' with argument '10'.
Reading option '/tmp/archive/e0c92fa0-dad5-11e7-8687-090dda95b1a4_10e1c990-dc70-11e7-888d-9f39ca0c79bc/1512775954465.m3u8' ... matched as output url.
Finished splitting the commandline.
Parsing a group of options: global .
Applying option y (overwrite output files) with argument 1.
1512775954592 - stderr: Applying option loglevel (set logging level) with argument debug.
Applying option dump (dump each input packet) with argument 1.
Applying option vsync (video sync method) with argument 0.
Successfully parsed a group of options.
Parsing a group of options: input url data:text/plain;base64,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.
Successfully parsed a group of options.
Opening an input file: data:text/plain;base64,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.
1512775954592 - stderr: [NULL @ 0x7f81fd000000] Opening 'data:text/plain;base64,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' for reading
1512775954593 - stderr: [data @ 0x7f81fca001a0] Content-type: text/plain
{"level":"info","time":"Dec 8, 2017 11:32 PM","message":"ffmpeg started"}
1512775954595 - stderr: [sdp @ 0x7f81fd000000] Format sdp probed with size=2048 and score=50
1512775954598 - stderr: [sdp @ 0x7f81fd000000] audio codec set to: opus
[sdp @ 0x7f81fd000000] audio samplerate set to: 48000
[sdp @ 0x7f81fd000000] audio channels set to: 2
1512775954639 - stderr: [sdp @ 0x7f81fd000000] video codec set to: h264
[sdp @ 0x7f81fd000000] RTP Packetization Mode: 1
[sdp @ 0x7f81fd000000] RTP Profile IDC: 42 Profile IOP: e0 Level: 1f
[udp @ 0x7f81fcb007e0] end receive buffer size reported is 65536
[udp @ 0x7f81fbe00180] end receive buffer size reported is 65536
[sdp @ 0x7f81fd000000] setting jitter buffer size to 500
[udp @ 0x7f81fbe00680] end receive buffer size reported is 65536
[udp @ 0x7f81fbe00740] end receive buffer size reported is 65536
[sdp @ 0x7f81fd000000] setting jitter buffer size to 500
[sdp @ 0x7f81fd000000] Before avformat_find_stream_info() pos: 479 bytes read:479 seeks:0 nb_streams:2
{"level":"info","time":"Dec 8, 2017 11:32 PM","message":"new active speaker","activePeer":"9f05d96a-9641-4c63-8f0e-486b98e48eb5"}
1512775954773 - stderr: [AVBSFContext @ 0x7f81fc8018e0] nal_unit_type: 7, nal_ref_idc: 3
[AVBSFContext @ 0x7f81fc8018e0] nal_unit_type: 8, nal_ref_idc: 3
[AVBSFContext @ 0x7f81fc8018e0] nal_unit_type: 5, nal_ref_idc: 3
1512775954774 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 7, nal_ref_idc: 3
1512775954774 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 8, nal_ref_idc: 3
1512775954774 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 5, nal_ref_idc: 3
1512775954774 - stderr: [h264 @ 0x7f8200000c00] Reinit context to 640x480, pix_fmt: yuv420p
1512775954801 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 1, nal_ref_idc: 3
1512775954939 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 7, nal_ref_idc: 3
1512775954939 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 8, nal_ref_idc: 3
1512775954940 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 5, nal_ref_idc: 3
1512775955003 - stderr: [h264 @ 0x7f8200000c00] nal_unit_type: 1, nal_ref_idc: 3
1512775955999 - stderr:     Last message repeated 3 times
[sdp @ 0x7f81fd000000] All info found
1512775955999 - stderr: [sdp @ 0x7f81fd000000] rfps: 29.750000 0.019566
[sdp @ 0x7f81fd000000] rfps: 29.833333 0.015263
[sdp @ 0x7f81fd000000] rfps: 29.916667 0.011503
1512775955999 - stderr: [sdp @ 0x7f81fd000000] rfps: 30.000000 0.008285
[sdp @ 0x7f81fd000000] rfps: 31.000000 0.011990
    Last message repeated 1 times
[sdp @ 0x7f81fd000000] rfps: 29.970030 0.009380
[sdp @ 0x7f81fd000000] After avformat_find_stream_info() pos: 479 bytes read:479 seeks:0 frames:98
1512775956000 - stderr: Input #0, sdp, from 'data:text/plain;base64,dj0wCm89LSAwIDAgSU4gSVA0IDEyNy4wLjAuMQpzPWUwYzkyZmEwLWRhZDUtMTFlNy04Njg3LTA5MGRkYTk1YjFhNCBmb29ib2FyCmM9SU4gSVA0IDEyNy4wLjAuMQptPWF1ZGlvIDIwMDAwIFJUUC9BVlBGIDEwMAphPXNlbmRyZWN2CmE9cnRjcC1tdXgKYT1ydHBtYXA6MTAwIG9wdXMvNDgwMDAvMgphPWZtdHA6MTAwIG1pbnB0aW1lPTEwOyB1c2VpbmJhbmRmZWM9MQptPXZpZGVvIDIwMDAyIFJUUC9BVlBGIDEwMQphPXNlbmRyZWN2CmE9cnRjcC1tdXgKYT1ydHBtYXA6MTAxIEgyNjQvOTAwMDAKYT1ydGNwLWZiOjEwMSBjY20gZmlyCmE9cnRjcC1mYjoxMDEgbmFjawphPXJ0Y3AtZmI6MTAxIG5hY2sgcGxpCmE9cnRjcC1mYjoxMDEgZ29vZy1yZW1iCmE9cnRjcC1mYjoxMDEgdHJhbnNwb3J0LWNjCmE9Zm10cDoxMDEgbGV2ZWwtYXN5bW1ldHJ5LWFsbG93ZWQ9MTtwYWNrZXRpemF0aW9uLW1vZGU9MTtwcm9maWxlLWxldmVsLWlkPTQyZTAxZgo=':
1512775956000 - stderr:   Metadata:
    title           : e0c92fa0-dad5-11e7-8687-090dda95b1a4 fooboar
  Duration: N/A, start: 0.000000, bitrate: N/A
    Stream #0:0, 70, 1/48000: Audio: opus, 48000 Hz, stereo, fltp1512775956000 - stderr:
    Stream #0:1, 28, 1/90000: Video: h264 (Constrained Baseline), 1 reference frame, yuv420p(progressive, left), 640x480, 0/1, 30 tbr, 90k tbn, 180k tbc
1512775956000 - stderr: Successfully opened the file.
Parsing a group of options: output url /tmp/archive/e0c92fa0-dad5-11e7-8687-090dda95b1a4_10e1c990-dc70-11e7-888d-9f39ca0c79bc/1512775954465.m3u8.
Applying option vcodec (force video codec ('copy' to copy stream)) with argument copy.
Applying option acodec (force audio codec ('copy' to copy stream)) with argument aac.
Applying option map (set input stream mapping) with argument 0:0,0:1.
Applying option map (set input stream mapping) with argument 0:1,0:1.
Applying option bsf:v (A comma-separated list of bitstream filters) with argument h264_mp4toannexb.
Successfully parsed a group of options.
1512775956000 - stderr: Opening an output file: /tmp/archive/e0c92fa0-dad5-11e7-8687-090dda95b1a4_10e1c990-dc70-11e7-888d-9f39ca0c79bc/1512775954465.m3u8.
1512775956000 - stderr: Successfully opened the file.
1512775956001 - stderr: [AVBSFContext @ 0x7f81fcb020a0] The input looks like it is Annex B already
Stream mapping:
1512775956001 - stderr:   Stream #0:0 -> #0:0 [sync #0:1] (opus (native) -> aac (native))
  Stream #0:1 -> #0:1 (copy)
Press [q] to stop, [?] for help
cur_dts is invalid (this is harmless if it occurs once at the start per stream)
stream #0:
  keyframe=1
  duration=0.000
  dts=0.000  pts=0.000
  size=82
1512775956001 - stderr: [SWR @ 0x7f820001e600] Using fltp internally between filters
1512775956002 - stderr: detected 8 logical cores
1512775956003 - stderr: [graph_0_in_0_0 @ 0x7f81fc90dfc0] Setting 'time_base' to value '1/48000'
[graph_0_in_0_0 @ 0x7f81fc90dfc0] Setting 'sample_rate' to value '48000'
1512775956003 - stderr: [graph_0_in_0_0 @ 0x7f81fc90dfc0] Setting 'sample_fmt' to value 'fltp'
[graph_0_in_0_0 @ 0x7f81fc90dfc0] Setting 'channel_layout' to value '0x3'
[graph_0_in_0_0 @ 0x7f81fc90dfc0] tb:1/48000 samplefmt:fltp samplerate:48000 chlayout:0x3
[format_out_0_0 @ 0x7f81fc914da0] Setting 'sample_fmts' to value 'fltp'
[format_out_0_0 @ 0x7f81fc914da0] Setting 'sample_rates' to value '96000|88200|64000|48000|44100|32000|24000|22050|16000|12000|11025|8000|7350'
1512775956004 - stderr: [AVFilterGraph @ 0x7f81fbd02200] query_formats: 4 queried, 9 merged, 0 already done, 0 delayed
1512775956007 - stderr: [hls @ 0x7f81fd80f000] Opening '/tmp/archive/e0c92fa0-dad5-11e7-8687-090dda95b1a4_10e1c990-dc70-11e7-888d-9f39ca0c79bc/15127759544650.ts' for writing
[file @ 0x7f81fbf01ee0] Setting default whitelist 'file,crypto'
1512775956007 - stderr: [mpegts @ 0x7f81fd877800] muxrate VBR, pcr every 9000 pkts, sdt every 2147483647, pat/pmt every 2147483647 pkts
Output #0, hls, to '/tmp/archive/e0c92fa0-dad5-11e7-8687-090dda95b1a4_10e1c990-dc70-11e7-888d-9f39ca0c79bc/1512775954465.m3u8':
  Metadata:
    title           : e0c92fa0-dad5-11e7-8687-090dda95b1a4 fooboar
    encoder         : Lavf57.83.100
1512775956007 - stderr:     Stream #0:0, 0, 1/90000: Audio: aac (LC), 48000 Hz, stereo, fltp, delay 1024, 128 kb/s
    Metadata:
      encoder         : Lavc57.107.100 aac
    Stream #0:1, 0, 1/90000: Video: h264 (Constrained Baseline), 1 reference frame, yuv420p(progressive, left), 640x480 (0x0), 0/1, q=2-31, 30 tbr, 90k tbn, 90k tbc
cur_dts is invalid (this is harmless if it occurs once at the start per stream)
1512775956007 - stderr:     Last message repeated 1 times
stream #0:
  keyframe=1
  duration=0.000
  dts=0.020  pts=0.020
  size=79
1512775956007 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
    Last message repeated 1 times
stream #0:
  keyframe=1
  duration=0.000
1512775956007 - stderr:   dts=0.040  pts=0.040
  size=75
1512775956014 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
stream #0:
  keyframe=1
  duration=0.000
  dts=0.0601512775956014 - stderr:   pts=0.060
  size=81
1512775956017 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
stream #0:
  keyframe=1
  duration=0.000
  dts=0.080  pts=0.080
  size=76
1512775956022 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
stream #0:
  keyframe=1
  duration=0.000
  dts=0.1001512775956022 - stderr:   pts=0.100
  size=79
1512775956023 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
stream #0:
  keyframe=1
1512775956023 - stderr:   duration=0.000
  dts=0.120  pts=0.120
  size=95
1512775956024 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
1512775956024 - stderr: stream #0:
  keyframe=1
  duration=0.000
  dts=0.140  pts=0.140
  size=93
1512775956025 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
stream #0:
  keyframe=1
1512775956025 - stderr:   duration=0.000
  dts=0.160  pts=0.160
  size=94
1512775956026 - stderr: cur_dts is invalid (this is harmless if it occurs once at the start per stream)
stream #1:
  keyframe=1
1512775956026 - stderr:   duration=0.000
  dts=N/A  pts=N/A
  size=992
[hls @ 0x7f81fd80f000] Timestamps are unset in a packet for stream 1. This is deprecated and will stop working in the future. Fix your code to set the timestamps properly
1512775956026 - stderr: stream #0:
  keyframe=1
  duration=0.000
1512775956026 - stderr:   dts=0.180  pts=0.180
  size=108
1512775956027 - stderr: stream #1:
  keyframe=0
  duration=0.000
1512775956027 - stderr:   dts=0.002  pts=0.002
  size=3047
[hls @ 0x7f81fd80f000] pkt->duration = 0, maybe the hls segment duration will not precise
stream #0:
  keyframe=1
  duration=0.000
  dts=0.200  pts=0.200
  size=89
1512775956060 - stderr: stream #0:
  keyframe=1
  duration=0.000
  dts=0.220  pts=0.220
  size=73
stream #0:
  keyframe=1
  duration=0.000
  dts=0.240  pts=0.240
  size=78
stream #0:
  keyframe=1
  duration=0.000
  dts=0.260  pts=0.260

Notice how the first frame for stream #1 (video) starts after a number of audio frames? Specifically, it starts at stream #0 dts/pts 0.18. In this situation, the a/v sync issue is pretty much unnoticeable, but with a bunch of repros, I've determined that the a/v sync offset is always the duration of however long audio frames were sent before the first video frame (sometimes seconds). I'm consistently starting the RTP streams only tens of ms apart, so I can't control for this variance on the input side.

After the initial audio frames come in, the first video frame has a dts/pts around 0. What ffmpeg setting would I use to adjust the timestamps accordingly? I don't care about losing the starting audio that doesn't have video, so any solution that would adjust the timestamps works.

Adding -use_wallclock_as_timestamps to the input fixes the issue when testing locally, but I won't call it an answer until I test with some network degradation. — artushin

Iñaki Baz Castillo Iñaki Baz Castillo · Accepted Answer · 2017-12-10T21:26:26

I don't know much about ffmpeg and I'm not sure it is capable of synchronizing separate audio&video RTP input streams. When a browser receives those audio/video RTP streams, there is no sync issue at all. In fact, mediasoup properly sets the RTP timestamp and RTCP ntp fields.

I suggest you asking in the ffmpeg mailing list.

Fixing A/V sync issues with RTP/RTCP sent from mediasoup

1 Answers