Transmux RTC to RTMP. #2252
Transmux RTC to RTMP. #2252
Conversation
Codecov Report
@@ Coverage Diff @@
## develop #2252 +/- ##
===========================================
- Coverage 55.55% 55.11% -0.45%
===========================================
Files 126 126
Lines 55494 55887 +393
===========================================
- Hits 30832 30801 -31
- Misses 24662 25086 +424 | Impacted Files | Coverage Δ | |' '| Impacted Files | Coverage Δ | | Continue to review full report at Codecov.
|
| @@ -1,9 +1,11 @@ | |||
| static const AVCodec * const codec_list[] = { | |||
| &ff_aac_encoder, | |||
| &ff_opus_encoder, | |||
| &ff_libfdk_aac_encoder, | |||
There was a problem hiding this comment.
FDK aac has a LICENSE issue, I remember it is GPL or something like that, SRS cannot use this LICENSE.
At most, SRS can only use LGPL and link it with so library. Of course, there is no problem with Apache, BSD, and MIT licenses.
TRANS_BY_GPT3
There was a problem hiding this comment.
Is that to say that the source code of fdk aac cannot be used to compile ffmpeg, and instead we need to use the .so library to compile ffmpeg, and then call fdk aac through ffmpeg's interface?
TRANS_BY_GPT3
| codec_ctx_->frame_size = frame_->nb_samples; | ||
| } | ||
|
|
||
| if (SrsAudioCodecIdAAC == codec_id_) { |
There was a problem hiding this comment.
This logic seems to be somewhat unreasonable. Does FFmpeg have a corresponding function to implement copying?
TRANS_BY_GPT3
| @@ -416,6 +432,10 @@ SrsAudioRecode::SrsAudioRecode(SrsAudioCodecId src_codec, SrsAudioCodecId dst_co | |||
| dec_ = NULL; | |||
| enc_ = NULL; | |||
| resample_ = NULL; | |||
|
|
|||
| fw_audio_dec = new SrsFileWriter(); | |||
| @@ -1273,6 +1275,157 @@ SrsRtpFrameBufferEnum SrsRtpJitterBuffer::InsertPacket(uint16_t seq, uint32_t ts | |||
| return buffer_state; | |||
| } | |||
|
|
|||
|
|
|||
| #if 1 | |||
| SrsRtpFrameBufferEnum SrsRtpJitterBuffer::InsertPacket2(const SrsRtpPacket2 *pkt, const char*nalu, int nalu_len, | |||
| } | ||
|
|
||
|
|
||
| //TODO:huping modify. |
There was a problem hiding this comment.
Why is it commented out?
TRANS_BY_GPT3
| @@ -498,6 +499,7 @@ class ISrsSourceBridger | |||
| virtual srs_error_t on_audio(SrsSharedPtrMessage* audio) = 0; | |||
| virtual srs_error_t on_video(SrsSharedPtrMessage* video) = 0; | |||
| virtual void on_unpublish() = 0; | |||
| virtual srs_error_t on_rtp(SrsRtpPacket2* pkt) = 0; | |||
There was a problem hiding this comment.
on_audio/on_video is the frame processing function for RTMP.
on_rtp is the packet processing function for RTP.
It is not ideal to have these two APIs together.
TRANS_BY_GPT3
|
Fixed by #2303 |
Hello srs maintainer:
The functionality of converting RTC to RTMP has been implemented based on the concept of srs birdger. This includes transcoding opus to aac and using jitter buffer to frame h264.
Issue:
#2200
TRANS_BY_GPT3