Resampling audio using libswresample from 48000 to 44100

Question

I'm trying to resample a decoded audio frame from 48KHz to 44.1KHz using the libswresample API. The code I have is the following:

// 'frame' is the original decoded audio frame
AVFrame *output_frame = av_frame_alloc();

// Without this, there is no sound at all at the output (PTS stuff I guess)
av_frame_copy_props(output_frame, frame);

output_frame->channel_layout = audioStream->codec->channel_layout;
output_frame->sample_rate = audioStream->codec->sample_rate;
output_frame->format = audioStream->codec->sample_fmt;

SwrContext *swr;
// Configure resampling context
swr = swr_alloc_set_opts(NULL,  // we're allocating a new context
                         AV_CH_LAYOUT_STEREO,  // out_ch_layout
                         AV_SAMPLE_FMT_FLTP,     // out_sample_fmt
                         44100,                // out_sample_rate
                         AV_CH_LAYOUT_STEREO,  // in_ch_layout
                         AV_SAMPLE_FMT_FLTP,   // in_sample_fmt
                         48000,                // in_sample_rate
                         0,                    // log_offset
                         NULL);                // log_ctx
// Initialize resampling context
swr_init(swr);

// Perform conversion
swr_convert_frame(swr, output_frame, frame);

// Close resampling context
swr_close(swr);
swr_free(&swr);
// Free the original frame and replace it with the new one
av_frame_unref(frame);
return output_frame;

With this code I'm able to hear the audio at the output but it is also noisy. From what I read, this code without the av_frame_copy_props() should be enough but it is not working for some reason. Any ideas?

EDIT: The input stream encodes the audio using AAC and the number of samples is 1024. But, after conversion, the number of samples is 925.

EDIT: I tried doing it in reverse. Since my app receives streams from any sources, some audio streams are 48KHz and some others 44.1KHz. So I tried resampling from 44.1 to 48 to avoid resampling loss. But now the frames has more than 1024 samples each one and the encoding fails.

EDIT: I tried using libavfilter instead with the following filter chain:

int init_filter_graph(AVStream *audio_st) {
// create new graph
filter_graph = avfilter_graph_alloc();
if (!filter_graph) {
    av_log(NULL, AV_LOG_ERROR, "unable to create filter graph: out of memory\n");
    return -1;
}

AVFilter *abuffer = avfilter_get_by_name("abuffer");
AVFilter *aformat = avfilter_get_by_name("aformat");
AVFilter *asetnsamples = avfilter_get_by_name("asetnsamples");
AVFilter *abuffersink = avfilter_get_by_name("abuffersink");

int err;
// create abuffer filter
AVCodecContext *avctx = audio_st->codec;
AVRational time_base = audio_st->time_base;
snprintf(strbuf, sizeof(strbuf),
         "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%" PRIx64,
         time_base.num, time_base.den, avctx->sample_rate,
         av_get_sample_fmt_name(avctx->sample_fmt),
         avctx->channel_layout);
fprintf(stderr, "abuffer: %s\n", strbuf);
err = avfilter_graph_create_filter(&abuffer_ctx, abuffer,
                                   NULL, strbuf, NULL, filter_graph);
if (err < 0) {
    av_log(NULL, AV_LOG_ERROR, "error initializing abuffer filter\n");
    return err;
}
// create aformat filter
snprintf(strbuf, sizeof(strbuf),
         "sample_fmts=%s:sample_rates=%d:channel_layouts=0x%" PRIx64,
         av_get_sample_fmt_name(AV_SAMPLE_FMT_FLTP), 44100,
         AV_CH_LAYOUT_STEREO);
fprintf(stderr, "aformat: %s\n", strbuf);
err = avfilter_graph_create_filter(&aformat_ctx, aformat,
                                   NULL, strbuf, NULL, filter_graph);
if (err < 0) {
    av_log(NULL, AV_LOG_ERROR, "unable to create aformat filter\n");
    return err;
}
// create asetnsamples filter
snprintf(strbuf, sizeof(strbuf),
         "n=1024:p=0");
fprintf(stderr, "asetnsamples: %s\n", strbuf);
err = avfilter_graph_create_filter(&asetnsamples_ctx, asetnsamples,
                                   NULL, strbuf, NULL, filter_graph);
if (err < 0) {
    av_log(NULL, AV_LOG_ERROR, "unable to create asetnsamples filter\n");
    return err;
}
// create abuffersink filter
err = avfilter_graph_create_filter(&abuffersink_ctx, abuffersink,
                                   NULL, NULL, NULL, filter_graph);
if (err < 0) {
    av_log(NULL, AV_LOG_ERROR, "unable to create aformat filter\n");
    return err;
}

// connect inputs and outputs
if (err >= 0) err = avfilter_link(abuffer_ctx, 0, aformat_ctx, 0);
if (err >= 0) err = avfilter_link(aformat_ctx, 0, asetnsamples_ctx, 0);
if (err >= 0) err = avfilter_link(asetnsamples_ctx, 0, abuffersink_ctx, 0);
if (err < 0) {
    av_log(NULL, AV_LOG_ERROR, "error connecting filters\n");
    return err;
}
err = avfilter_graph_config(filter_graph, NULL);
if (err < 0) {
    av_log(NULL, AV_LOG_ERROR, "error configuring the filter graph\n");
    return err;
}
return 0;
}

Now the resulting frame has 1024 samples but the audio is still choppy.

Ruben Sanchez Castellano Ruben Sanchez Castellano · Accepted Answer · 2017-08-09T15:49:32

Finally, I get rid of this issue with the solution from here.

This is the code of the filter creation for my setup (resampling to 44.1KHz)

AVFilterGraph *filter_graph = NULL;
AVFilterContext *buffersrc_ctx = NULL;
AVFilterContext *buffersink_ctx = NULL;
QString filter_description = "aresample=44100,aformat=sample_fmts=fltp:channel_layouts=stereo,asetnsamples=n=1024:p=0";
/**
* Initialize conversion filter */
int initialize_audio_filter(AVStream *inputStream) {
char args[512];
int ret;
AVFilter *buffersrc = avfilter_get_by_name("abuffer");
AVFilter *buffersink = avfilter_get_by_name("abuffersink");
AVFilterInOut *outputs = avfilter_inout_alloc();
AVFilterInOut *inputs = avfilter_inout_alloc();
filter_graph = avfilter_graph_alloc();
const enum AVSampleFormat out_sample_fmts[] = {AV_SAMPLE_FMT_FLTP, AV_SAMPLE_FMT_NONE};
const int64_t out_channel_layouts[] = {AV_CH_LAYOUT_STEREO, -1};
const int out_sample_rates[] = {44100, -1};

snprintf(args, sizeof(args), "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%" PRIx64,
         inputStream->codec->time_base.num, inputStream->codec->time_base.den,
         inputStream->codec->sample_rate,
         av_get_sample_fmt_name(inputStream->codec->sample_fmt),
         inputStream->codec->channel_layout);
ret = avfilter_graph_create_filter(&buffersrc_ctx, buffersrc, "in", args, NULL, filter_graph);

if (ret < 0) {
    svsCritical("", QString("Could not create filter graph, error: %1").arg(svsAvErrorToFormattedString(ret)))
    return -1;
}

ret = avfilter_graph_create_filter(&buffersink_ctx, buffersink, "out", NULL, NULL, filter_graph);

if (ret < 0) {
    svsCritical("", QString("Cannot create buffer sink, error: %1").arg(svsAvErrorToFormattedString(ret)))
    return ret;
}

ret = av_opt_set_int_list(buffersink_ctx, "sample_fmts", out_sample_fmts, -1,
                          AV_OPT_SEARCH_CHILDREN);

if (ret < 0) {
    svsCritical("", QString("Cannot set output sample format, error: %1").arg(svsAvErrorToFormattedString(ret)))
    return ret;
}

ret = av_opt_set_int_list(buffersink_ctx, "channel_layouts", out_channel_layouts, -1,
                          AV_OPT_SEARCH_CHILDREN);

if (ret < 0) {
    svsCritical("", QString("Cannot set output channel layout, error: %1").arg(svsAvErrorToFormattedString(ret)))
    return ret;
}

ret = av_opt_set_int_list(buffersink_ctx, "sample_rates", out_sample_rates, -1,
                          AV_OPT_SEARCH_CHILDREN);

if (ret < 0) {
    svsCritical("", QString("Cannot set output sample rate, error: %1").arg(svsAvErrorToFormattedString(ret)))
    return ret;
}

/* Endpoints for the filter graph. */
outputs -> name = av_strdup("in");
outputs -> filter_ctx = buffersrc_ctx;
outputs -> pad_idx = 0;
outputs -> next = NULL;
/* Endpoints for the filter graph. */
inputs -> name = av_strdup("out");
inputs -> filter_ctx = buffersink_ctx;
inputs -> pad_idx = 0;
inputs -> next = NULL;

if ((ret = avfilter_graph_parse_ptr(filter_graph, filter_description.toStdString().c_str(), &inputs, &outputs, NULL)) < 0) {
    svsCritical("", QString("Could not add the filter to graph, error: %1").arg(svsAvErrorToFormattedString(ret)))
}

if ((ret = avfilter_graph_config(filter_graph, NULL)) < 0) {
    svsCritical("", QString("Could not configure the graph, error: %1").arg(svsAvErrorToFormattedString(ret)))
}

/* Print summary of the sink buffer
 * Note: args buffer is reused to store channel layout string */
AVFilterLink *outlink = buffersink_ctx->inputs[0];
av_get_channel_layout_string(args, sizeof(args), -1, outlink->channel_layout);
svsInfo("", QString::asprintf("Output: srate:%dHz fmt:%s chlayout:%s\n",
                              (int) outlink->sample_rate,
                              (char *) av_x_if_null(av_get_sample_fmt_name((AVSampleFormat) outlink->format), "?"),
                              args))
return 0;
}

And the filter usage:

AVFrame* resampleAudio(const QString& key, AVFrame *frame) {

    /* Push the decoded frame into the filtergraph */
    qint32 ret;
    ret = av_buffersrc_add_frame_flags(buffersrc_ctx, frame, AV_BUFFERSRC_FLAG_KEEP_REF);
    if(ret < 0) {
        svsWarning(key, QString("Error adding frame to buffer: %1").arg(svsAvErrorToFormattedString(ret)))
        // Delete input frame and return null
        av_frame_unref(frame);
        return nullptr;
    }

    AVFrame *resampled_frame = av_frame_alloc();

    /* Pull filtered frames from the filtergraph */
    ret = av_buffersink_get_frame(buffersink_ctx, resampled_frame);

    /* Set the timestamp on the resampled frame */
    resampled_frame->best_effort_timestamp = resampled_frame->pts;

    if(ret < 0) {
        // This is very common. For 48KHz -> 44.1KHz for some input frames the
        // filter has not data enough to generate another one.
        av_frame_unref(frame);
        av_frame_unref(resampled_frame);
        return nullptr;
    }
    av_frame_unref(frame);
    return resampled_frame;
}

Its important to set the best_effort_timestamp on the resampled frame to make it work. But the PTS of this frame is set by the filter.

Resampling audio using libswresample from 48000 to 44100

2 Answers