gstreamer/subprojects/gst-plugins-bad/ext/lc3/gstlc3enc.c

425 lines
13 KiB
C

/* GStreamer LC3 Bluetooth LE audio encoder
* Copyright (C) 2023 Asymptotic Inc. <taruntej@asymptotic.io>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 51 Franklin Street, Suite 500,
* Boston, MA 02110-1335, USA.
*/
/**
* SECTION:element-lc3enc
*
* The lc3enc element encodes raw audio using the Low Complexity Communication
* Codec (LC3).
*
* ## Example pipeline
* |[
* gst-launch-1.0 audiotestsrc ! lc3enc ! audio/x-lc3,channels=2,rate=48000,frame-duration-us=10000 !\
* filesink location=audio.lc3
* ]|
*
* Encodes a sine wave into LC3 format using the config params frame-duration-us
* specified by the caps downstream and save it to file audio.lc3
*
* Since: 1.24
*/
#include <stdlib.h>
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include <gst/gst.h>
#include <gst/audio/gstaudioencoder.h>
#include "gstlc3common.h"
#include "gstlc3enc.h"
GST_DEBUG_CATEGORY_STATIC (gst_lc3_enc_debug_category);
#define GST_CAT_DEFAULT gst_lc3_enc_debug_category
#define parent_class gst_lc3_enc_parent_class
G_DEFINE_TYPE (GstLc3Enc, gst_lc3_enc, GST_TYPE_AUDIO_ENCODER);
GST_ELEMENT_REGISTER_DEFINE (lc3enc, "lc3enc", GST_RANK_NONE, GST_TYPE_LC3_ENC);
static gboolean gst_lc3_enc_start (GstAudioEncoder * encoder);
static gboolean gst_lc3_enc_stop (GstAudioEncoder * encoder);
static gboolean gst_lc3_enc_set_format (GstAudioEncoder * encoder,
GstAudioInfo * info);
static GstFlowReturn gst_lc3_enc_handle_frame (GstAudioEncoder * encoder,
GstBuffer * buffer);
#define DEFAULT_BITRATE_PER_CHANNEL 160000
static GstStaticPadTemplate gst_lc3_enc_src_template =
GST_STATIC_PAD_TEMPLATE ("src",
GST_PAD_SRC,
GST_PAD_ALWAYS,
GST_STATIC_CAPS ("audio/x-lc3, "
"rate = (int) { " SAMPLE_RATES " }, "
"channels = (int) [1, MAX], "
"frame-bytes = (int) [" FRAME_BYTES_RANGE "], "
"frame-duration-us = (int) { " FRAME_DURATIONS "}, "
"framed=(boolean) true")
);
static GstStaticPadTemplate gst_lc3_enc_sink_template =
GST_STATIC_PAD_TEMPLATE ("sink",
GST_PAD_SINK,
GST_PAD_ALWAYS,
GST_STATIC_CAPS ("audio/x-raw, "
"format = " FORMAT ", "
"rate = (int) { " SAMPLE_RATES " }, channels = (int) [1, MAX]")
);
static void
gst_lc3_enc_class_init (GstLc3EncClass * klass)
{
GstAudioEncoderClass *audio_encoder_class = GST_AUDIO_ENCODER_CLASS (klass);
audio_encoder_class->start = GST_DEBUG_FUNCPTR (gst_lc3_enc_start);
audio_encoder_class->stop = GST_DEBUG_FUNCPTR (gst_lc3_enc_stop);
audio_encoder_class->set_format = GST_DEBUG_FUNCPTR (gst_lc3_enc_set_format);
audio_encoder_class->handle_frame =
GST_DEBUG_FUNCPTR (gst_lc3_enc_handle_frame);
gst_element_class_add_static_pad_template (GST_ELEMENT_CLASS (klass),
&gst_lc3_enc_src_template);
gst_element_class_add_static_pad_template (GST_ELEMENT_CLASS (klass),
&gst_lc3_enc_sink_template);
gst_element_class_set_static_metadata (GST_ELEMENT_CLASS (klass),
"LC3 Bluetooth Audio encoder", "Codec/Encoder/Audio",
"Encodes a raw audio stream to LC3",
"Taruntej Kanakamalla <taruntej@asymptotic.io>");
GST_DEBUG_CATEGORY_INIT (gst_lc3_enc_debug_category, "lc3enc", 0,
"debug category for lc3enc element");
}
static void
gst_lc3_enc_init (GstLc3Enc * lc3_enc)
{
}
static gboolean
gst_lc3_enc_start (GstAudioEncoder * encoder)
{
GstLc3Enc *lc3_enc = GST_LC3_ENC (encoder);
lc3_enc->enc_ch = NULL;
lc3_enc->frame_bytes = 0;
/* Set to true at the start of processing */
lc3_enc->first_frame = TRUE;
lc3_enc->pending_bytes = 0;
return TRUE;
}
static gboolean
gst_lc3_enc_stop (GstAudioEncoder * encoder)
{
GstLc3Enc *lc3_enc = GST_LC3_ENC (encoder);
if (lc3_enc->enc_ch != NULL) {
for (int ich = 0; ich < lc3_enc->channels; ich++) {
g_free (lc3_enc->enc_ch[ich]);
lc3_enc->enc_ch[ich] = NULL;
}
g_free (lc3_enc->enc_ch);
lc3_enc->enc_ch = NULL;
}
return TRUE;
}
static gboolean
gst_lc3_enc_set_format (GstAudioEncoder * encoder, GstAudioInfo * info)
{
GstLc3Enc *lc3_enc = GST_LC3_ENC (encoder);
GstCaps *caps = NULL, *filter_caps = NULL;
GstCaps *output_caps = NULL;
GstStructure *s;
GstClockTime latency;
lc3_enc->bpf = GST_AUDIO_INFO_BPF (info);
switch (GST_AUDIO_INFO_FORMAT (info)) {
case GST_AUDIO_FORMAT_S16LE:
lc3_enc->format = LC3_PCM_FORMAT_S16;
break;
case GST_AUDIO_FORMAT_S24LE:
lc3_enc->format = LC3_PCM_FORMAT_S24_3LE;
break;
case GST_AUDIO_FORMAT_F32:
lc3_enc->format = LC3_PCM_FORMAT_FLOAT;
break;
case GST_AUDIO_FORMAT_S24_32LE:
default:
lc3_enc->format = LC3_PCM_FORMAT_S24;
break;
}
caps = gst_pad_get_allowed_caps (GST_AUDIO_ENCODER_SRC_PAD (lc3_enc));
if (caps == NULL)
caps = gst_static_pad_template_get_caps (&gst_lc3_enc_src_template);
else if (gst_caps_is_empty (caps))
goto failure;
filter_caps = gst_caps_new_simple ("audio/x-lc3", "rate", G_TYPE_INT,
GST_AUDIO_INFO_RATE (info), "channels", G_TYPE_INT,
GST_AUDIO_INFO_CHANNELS (info), NULL);
output_caps = gst_caps_intersect (caps, filter_caps);
if (output_caps == NULL || gst_caps_is_empty (output_caps)) {
GST_WARNING_OBJECT (lc3_enc,
"Couldn't negotiate filter caps %" GST_PTR_FORMAT
" and allowed output caps %" GST_PTR_FORMAT, filter_caps, caps);
goto failure;
}
gst_caps_unref (filter_caps);
filter_caps = NULL;
gst_caps_unref (caps);
caps = NULL;
GST_DEBUG_OBJECT (lc3_enc, "fixating caps %" GST_PTR_FORMAT, output_caps);
output_caps = gst_caps_truncate (output_caps);
GST_DEBUG_OBJECT (lc3_enc, "truncated caps %" GST_PTR_FORMAT, output_caps);
s = gst_caps_get_structure (output_caps, 0);
gst_structure_get_int (s, "rate", &lc3_enc->rate);
gst_structure_get_int (s, "channels", &lc3_enc->channels);
gst_structure_get_int (s, "frame-bytes", &lc3_enc->frame_bytes);
if (gst_structure_fixate_field (s, "frame-duration-us")) {
gst_structure_get_int (s, "frame-duration-us", &lc3_enc->frame_duration_us);
} else {
lc3_enc->frame_duration_us = FRAME_DURATION_10000US;
GST_INFO_OBJECT (lc3_enc, "Frame duration not fixed, setting to %d",
lc3_enc->frame_duration_us);
gst_caps_set_simple (output_caps, "frame-duration-us", G_TYPE_INT,
lc3_enc->frame_duration_us, NULL);
}
if (lc3_enc->frame_bytes == 0) {
/* fixate_field() is always setting the frame_bytes to 20 which is not desired
* since we can get the value using frame duration and default bitrate
* compute the frame bytes and set the value to the caps
*/
lc3_enc->frame_bytes = lc3_frame_bytes (lc3_enc->frame_duration_us,
DEFAULT_BITRATE_PER_CHANNEL);
GST_INFO_OBJECT (lc3_enc, "frame bytes computed %d using duration %d",
lc3_enc->frame_bytes, lc3_enc->frame_duration_us);
gst_caps_set_simple (output_caps, "frame-bytes", G_TYPE_INT,
lc3_enc->frame_bytes, NULL);
}
GST_INFO_OBJECT (lc3_enc, "output caps %" GST_PTR_FORMAT, output_caps);
lc3_enc->frame_samples =
lc3_frame_samples (lc3_enc->frame_duration_us, lc3_enc->rate);
gst_audio_encoder_set_frame_samples_min (encoder, lc3_enc->frame_samples);
gst_audio_encoder_set_frame_samples_max (encoder, lc3_enc->frame_samples);
gst_audio_encoder_set_frame_max (encoder, 1);
latency =
gst_util_uint64_scale_int (lc3_enc->frame_samples, GST_SECOND,
lc3_enc->rate);
gst_audio_encoder_set_latency (encoder, latency, latency);
/* Free the encoder handles if it was initialised previously */
if (lc3_enc->enc_ch != NULL) {
for (int ich = 0; ich < lc3_enc->channels; ich++) {
g_free (lc3_enc->enc_ch[ich]);
lc3_enc->enc_ch[ich] = NULL;
}
g_free (lc3_enc->enc_ch);
lc3_enc->enc_ch = NULL;
}
lc3_enc->enc_ch =
(lc3_encoder_t *) g_malloc (sizeof (lc3_encoder_t) * lc3_enc->channels);
for (guint8 i = 0; i < lc3_enc->channels; i++) {
/* The encoder can resample for us. But we leave the resampling to
* happen before encoding explicitly for now. So pass the same sample rate
* for sr_hz and sr_pcm_hz
*/
lc3_enc->enc_ch[i] =
lc3_setup_encoder (lc3_enc->frame_duration_us, lc3_enc->rate,
lc3_enc->rate, g_malloc (lc3_encoder_size (lc3_enc->frame_duration_us,
lc3_enc->rate)));
if (lc3_enc->enc_ch[i] == NULL) {
GST_ERROR_OBJECT (lc3_enc,
"Failed to create encoder handle for channel %" G_GUINT32_FORMAT, i);
goto failure;
}
}
if (!gst_audio_encoder_set_output_format (encoder, output_caps))
goto failure;
gst_caps_unref (output_caps);
return gst_audio_encoder_negotiate (encoder);
failure:
if (output_caps)
gst_caps_unref (output_caps);
if (caps)
gst_caps_unref (caps);
if (filter_caps)
gst_caps_unref (filter_caps);
return FALSE;
}
static GstFlowReturn
gst_lc3_enc_handle_frame (GstAudioEncoder * encoder, GstBuffer * buffer)
{
GstLc3Enc *lc3_enc = GST_LC3_ENC (encoder);
GstMapInfo in_map = GST_MAP_INFO_INIT, out_map = GST_MAP_INFO_INIT;
GstBuffer *outbuf = NULL;
guint samplesize, stride, req_samples, req_bytes, frame_bytes;
guint8 *pcm_in;
gint ret = -1;
guint64 trim_start = 0, trim_end = 0;
if (buffer == NULL && !lc3_enc->pending_bytes)
return GST_FLOW_OK;
if (G_UNLIKELY (lc3_enc->channels == 0))
return GST_FLOW_ERROR;
if (buffer && !gst_buffer_map (buffer, &in_map, GST_MAP_READ))
goto map_failed;
GST_TRACE_OBJECT (lc3_enc,
"encoding %" G_GSIZE_FORMAT " frame samples of %" G_GSIZE_FORMAT
" bytes", in_map.size / lc3_enc->bpf, in_map.size);
frame_bytes = lc3_enc->frame_bytes;
/* allocate frame_bytes for each channel in the output buffer */
outbuf =
gst_audio_encoder_allocate_output_buffer (encoder,
frame_bytes * lc3_enc->channels);
if (outbuf == NULL)
goto no_buffer;
if (!gst_buffer_map (outbuf, &out_map, GST_MAP_WRITE))
goto map_failed;
stride = lc3_enc->channels;
samplesize = lc3_enc->bpf / lc3_enc->channels;
/* Calculate the expected bytes */
req_samples = lc3_enc->frame_samples;
req_bytes = req_samples * lc3_enc->bpf;
if (lc3_enc->first_frame) {
/* LC3 encoder introduces extra samples as a part of the
* algorithmic delay at the beginning of the frame
*/
lc3_enc->pending_bytes =
lc3_enc->bpf * lc3_delay_samples (lc3_enc->frame_duration_us,
lc3_enc->rate);
/* trim start 'delay_samples' bytes for the first frame */
trim_start = lc3_enc->pending_bytes / lc3_enc->bpf;
lc3_enc->first_frame = FALSE;
}
if (in_map.size < req_bytes) {
/* update the pending bytes and trim_end */
if (in_map.size + lc3_enc->pending_bytes > req_bytes) {
lc3_enc->pending_bytes = in_map.size + lc3_enc->pending_bytes - req_bytes;
} else {
trim_end =
(req_bytes - in_map.size - lc3_enc->pending_bytes) / lc3_enc->bpf;
lc3_enc->pending_bytes = 0;
}
/* The encoder always expects fixed number of bytes in the input
* If we get less bytes than req_bytes, most likely in the last iteration,
* add zero-padding bytes at the end
*/
pcm_in = (guint8 *) g_malloc0 (req_bytes);
if (in_map.size && in_map.data)
memcpy (pcm_in, in_map.data, in_map.size);
} else {
pcm_in = in_map.data;
}
if (trim_start || trim_end) {
GST_TRACE_OBJECT (lc3_enc,
"Adding trim-start %" G_GUINT64_FORMAT " trim-end %" G_GUINT64_FORMAT,
trim_start, trim_end);
gst_buffer_add_audio_clipping_meta (outbuf, GST_FORMAT_DEFAULT, trim_start,
trim_end);
}
for (guint8 ch = 0; ch < lc3_enc->channels; ch++) {
ret = lc3_encode (lc3_enc->enc_ch[ch], lc3_enc->format,
pcm_in + (ch * samplesize), stride, frame_bytes,
out_map.data + (ch * frame_bytes));
if (ret < 0) {
GST_WARNING_OBJECT (lc3_enc,
"encoding error: invalid enc handle or frame_bytes");
break;
}
}
if (in_map.size < req_bytes)
g_free (pcm_in);
gst_buffer_unmap (outbuf, &out_map);
if (buffer)
gst_buffer_unmap (buffer, &in_map);
if (ret < 0)
return GST_FLOW_ERROR;
return gst_audio_encoder_finish_frame (encoder, outbuf, req_samples);
no_buffer:
{
if (buffer)
gst_buffer_unmap (buffer, &in_map);
GST_ELEMENT_ERROR (lc3_enc, STREAM, FAILED, (NULL),
("Could not allocate output buffer"));
return GST_FLOW_ERROR;
}
map_failed:
{
if (buffer)
gst_buffer_unmap (buffer, &in_map);
GST_ELEMENT_ERROR (lc3_enc, STREAM, FAILED, (NULL),
("Failed to get the buffer memory map"));
return GST_FLOW_ERROR;
}
}