external/srs
1
0
Fork 0
mirror of https://github.com/ossrs/srs.git synced 2025-03-09 15:49:59 +00:00
Code Issues Releases Wiki Activity

RTC: Refine FFmpeg opus audio noisy issue. v5.0.197 v6.0.97 (#3852)

Browse source 
### Description

When converting between AAC and Opus formats (aac2opus or opus2aac), the
`av_frame_get_buffer` API is frequently called.

### Objective

The goal is to optimize the code logic and reduce the frequent
allocation and deallocation of memory.

In the case of aac2opus, av_frame_get_buffer is still frequently called.
In the case of opus2aac, the goal is to avoid calling
av_frame_get_buffer and reduce memory allocations.

### Additional Note

Before calling the `av_audio_fifo_read` API, use
`av_frame_make_writable` to check if the frame is writable. If it is not
writable, create a new frame.

---------

Co-authored-by: john <hondaxiao@tencent.com>
This commit is contained in:
chundonglinlin 2023-11-04 16:21:44 +08:00 • committed by GitHub
parent 4a100616fc
commit e7b629cd39
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
10 changed files with 27 additions and 24 deletions
Download patch file Download diff file Expand all files Collapse all files

20
trunk/src/app/srs_app_rtc_codec.cpp
View file

@ -260,6 +260,13 @@ srs_error_t SrsAudioTranscoder::init_enc(SrsAudioCodecId dst_codec, int dst_chan
if (!enc_frame_) {
return srs_error_new(ERROR_RTC_RTP_MUXER, "Could not allocate audio encode in frame");
}
enc_frame_->format = enc_->sample_fmt;
enc_frame_->nb_samples = enc_->frame_size;
enc_frame_->channel_layout = enc_->channel_layout;
if (av_frame_get_buffer(enc_frame_, 0) < 0) {
return srs_error_new(ERROR_RTC_RTP_MUXER, "Could not get audio frame buffer");
}
enc_packet_ = av_packet_alloc();
if (!enc_packet_) {
@ -381,26 +388,21 @@ srs_error_t SrsAudioTranscoder::encode(std::vector<SrsAudioFrame*> &pkts)
}
while (av_audio_fifo_size(fifo_) >= enc_->frame_size) {
enc_frame_->format = enc_->sample_fmt;
enc_frame_->nb_samples = enc_->frame_size;
enc_frame_->channel_layout = enc_->channel_layout;
if (av_frame_get_buffer(enc_frame_, 0) < 0) {
av_frame_free(&enc_frame_);
return srs_error_new(ERROR_RTC_RTP_MUXER, "Could not get audio frame buffer");
// make sure the frame is writable
if (av_frame_make_writable(enc_frame_) < 0) {
return srs_error_new(ERROR_RTC_RTP_MUXER, "Could not make writable frame");
}
/* Read as many samples from the FIFO buffer as required to fill the frame.
* The samples are stored in the frame temporarily. */
if (av_audio_fifo_read(fifo_, (void **)enc_frame_->data, enc_->frame_size) < enc_->frame_size) {
av_frame_free(&enc_frame_);
return srs_error_new(ERROR_RTC_RTP_MUXER, "Could not read data from FIFO");
}
/* send the frame for encoding */
enc_frame_->pts = next_out_pts_;
next_out_pts_ += enc_->frame_size;
int error = avcodec_send_frame(enc_, enc_frame_);
av_frame_unref(enc_frame_);
if (error < 0) {
return srs_error_new(ERROR_RTC_RTP_MUXER, "Error sending the frame to the encoder(%d,%s)", error,
av_make_error_string(err_buf, AV_ERROR_MAX_STRING_SIZE, error));