FFmpeg programming study NOTE 2: audio re-sampling, ffmpeg Study Notes

FFmpeg programming study NOTE 2: audio re-sampling, ffmpeg Study Notes

The core function swr_convert of ffmpeg for Implementing Audio re-sampling is very powerful. However, the comment of ffmpeg is too simple and such problems often occur in applications, in fact, swr_convert is also applied after the first execution of swr_convert in the loop of reading data> resampling> encoding data to check whether there is any data, if there is a need to write it into the FIFO, it will be used again next time, this is particularly important in transcoding and high conversion from low sampling rate.

The following simple code is taken from my own test program.

const int frame_size = FFMIN(fifo_size, m_Opt->encode_pCodecCtx->frame_size);if ((ret = av_audio_fifo_read(m_fifo, (void **)m_fifo_samples_array, frame_size)) < frame_size) {fprintf(stderr, "Could not read data from FIFO\n");return AVERROR_EXIT;}int out_samples = av_rescale_rnd(swr_get_delay(m_Opt->out_resample_context, 48000) + 1536, 44100, 48000, AV_ROUND_UP);int conver_samples= swr_convert(m_Opt->out_resample_context, m_fifo_conver_samples_array, frame_size,(const uint8_t **)m_fifo_samples_array, frame_size);ret = av_audio_fifo_size(m_conver_fifo);if ((ret = av_audio_fifo_realloc(m_conver_fifo, av_audio_fifo_size(m_conver_fifo) + frame_size)) < 0) {fprintf(stderr, "Could not reallocate FIFO\n");return ret;}ret = av_audio_fifo_write(m_conver_fifo, (void **)m_fifo_conver_samples_array, conver_samples);fifo_size = av_audio_fifo_size(m_conver_fifo);if ((ret = av_audio_fifo_read(m_conver_fifo, (void **)m_Opt->out_samples_array, frame_size)) < frame_size) {fprintf(stderr, "Could not read data from FIFO\n");return AVERROR_EXIT;}ret = Encode_audio(ret);int ret1 = 0;while ((ret1 = swr_convert(m_Opt->out_resample_context, m_fifo_conver_samples_array, frame_size, NULL, 0)) > 0){if ((ret = av_audio_fifo_realloc(m_conver_fifo, av_audio_fifo_size(m_conver_fifo) + ret1)) < 0) {fprintf(stderr, "Could not reallocate FIFO\n");return ret;}ret = av_audio_fifo_write(m_conver_fifo, (void **)m_fifo_conver_samples_array, ret1);fifo_size = av_audio_fifo_size(m_conver_fifo);if (fifo_size > m_Opt->encode_pFrame->nb_samples){if ((ret = av_audio_fifo_read(m_conver_fifo, (void **)m_Opt->out_samples_array, frame_size)) < frame_size) {fprintf(stderr, "Could not read data from FIFO\n");return AVERROR_EXIT;}ret = Encode_audio(ret);}

Because it is a test code, there are some unnecessary steps in the middle, mainly for debugging.

Focus:

    if ((r = swr_convert(swr_ctx, output, output_nb_samples,(const uint8_t**)input, nb_samples)) < 0)        return -1;    while ((r = swr_convert(swr_ctx, output, output_nb_samples, NULL, 0)) > 0) {     }

Copyright Disclaimer: This article is an original article by the blogger and cannot be reproduced without the permission of the blogger.