[FFmpeg-cvslog] avcodec/g723_1: add support for stereo files
Paul B Mahol
git at videolan.org
Tue Dec 18 20:02:26 EET 2018
ffmpeg | branch: master | Paul B Mahol <onemda at gmail.com> | Sat Dec 15 13:20:43 2018 +0100| [62dbcb7ddf106ce43018d3087f8fc49286a1fbc5] | committer: Paul B Mahol
avcodec/g723_1: add support for stereo files
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=62dbcb7ddf106ce43018d3087f8fc49286a1fbc5
---
libavcodec/g723_1.h | 13 +++++++-----
libavcodec/g723_1dec.c | 54 +++++++++++++++++++++++++++++---------------------
libavcodec/g723_1enc.c | 14 +++++++------
3 files changed, 47 insertions(+), 34 deletions(-)
diff --git a/libavcodec/g723_1.h b/libavcodec/g723_1.h
index f833af01c6..d60d481e67 100644
--- a/libavcodec/g723_1.h
+++ b/libavcodec/g723_1.h
@@ -116,9 +116,7 @@ typedef struct FCBParam {
int pulse_sign[PULSE_MAX];
} FCBParam;
-typedef struct g723_1_context {
- AVClass *class;
-
+typedef struct G723_1_ChannelContext {
G723_1_Subframe subframe[4];
enum FrameType cur_frame_type;
enum FrameType past_frame_type;
@@ -144,8 +142,6 @@ typedef struct g723_1_context {
int reflection_coef;
int pf_gain; ///< formant postfilter
///< gain scaling unit memory
- int postfilter;
-
int16_t audio[FRAME_LEN + LPC_ORDER + PITCH_MAX + 4];
/* encoder */
@@ -158,6 +154,13 @@ typedef struct g723_1_context {
int16_t perf_iir_mem[LPC_ORDER]; ///< and iir memories
int16_t harmonic_mem[PITCH_MAX];
+} G723_1_ChannelContext;
+
+typedef struct G723_1_Context {
+ AVClass *class;
+ int postfilter;
+
+ G723_1_ChannelContext ch[2];
} G723_1_Context;
diff --git a/libavcodec/g723_1dec.c b/libavcodec/g723_1dec.c
index ab952ec66d..798d464c71 100644
--- a/libavcodec/g723_1dec.c
+++ b/libavcodec/g723_1dec.c
@@ -42,12 +42,16 @@
static av_cold int g723_1_decode_init(AVCodecContext *avctx)
{
- G723_1_Context *p = avctx->priv_data;
+ G723_1_Context *s = avctx->priv_data;
+ G723_1_ChannelContext *p = &s->ch[0];
- avctx->channel_layout = AV_CH_LAYOUT_MONO;
- avctx->sample_fmt = AV_SAMPLE_FMT_S16;
- avctx->channels = 1;
- p->pf_gain = 1 << 12;
+ avctx->sample_fmt = AV_SAMPLE_FMT_S16P;
+ if (avctx->channels < 1 || avctx->channels > 2) {
+ av_log(avctx, AV_LOG_ERROR, "Only mono and stereo are supported (requested channels: %d).\n", avctx->channels);
+ return AVERROR(EINVAL);
+ }
+ avctx->channel_layout = avctx->channels == 1 ? AV_CH_LAYOUT_MONO : AV_CH_LAYOUT_STEREO;
+ p->pf_gain = 1 << 12;
memcpy(p->prev_lsp, dc_lsp, LPC_ORDER * sizeof(*p->prev_lsp));
memcpy(p->sid_lsp, dc_lsp, LPC_ORDER * sizeof(*p->sid_lsp));
@@ -65,7 +69,7 @@ static av_cold int g723_1_decode_init(AVCodecContext *avctx)
* @param buf pointer to the input buffer
* @param buf_size size of the input buffer
*/
-static int unpack_bitstream(G723_1_Context *p, const uint8_t *buf,
+static int unpack_bitstream(G723_1_ChannelContext *p, const uint8_t *buf,
int buf_size)
{
GetBitContext gb;
@@ -344,7 +348,7 @@ static void comp_ppf_gains(int lag, PPFParam *ppf, enum Rate cur_rate,
* @param ppf pitch postfilter parameters
* @param cur_rate current bitrate
*/
-static void comp_ppf_coeff(G723_1_Context *p, int offset, int pitch_lag,
+static void comp_ppf_coeff(G723_1_ChannelContext *p, int offset, int pitch_lag,
PPFParam *ppf, enum Rate cur_rate)
{
@@ -430,7 +434,7 @@ static void comp_ppf_coeff(G723_1_Context *p, int offset, int pitch_lag,
*
* @return residual interpolation index if voiced, 0 otherwise
*/
-static int comp_interp_index(G723_1_Context *p, int pitch_lag,
+static int comp_interp_index(G723_1_ChannelContext *p, int pitch_lag,
int *exc_eng, int *scale)
{
int offset = PITCH_MAX + 2 * SUBFRAME_LEN;
@@ -529,7 +533,7 @@ static void residual_interp(int16_t *buf, int16_t *out, int lag,
* @param buf postfiltered output vector
* @param energy input energy coefficient
*/
-static void gain_scale(G723_1_Context *p, int16_t * buf, int energy)
+static void gain_scale(G723_1_ChannelContext *p, int16_t * buf, int energy)
{
int num, denom, gain, bits1, bits2;
int i;
@@ -572,7 +576,7 @@ static void gain_scale(G723_1_Context *p, int16_t * buf, int energy)
* @param buf input buffer
* @param dst output buffer
*/
-static void formant_postfilter(G723_1_Context *p, int16_t *lpc,
+static void formant_postfilter(G723_1_ChannelContext *p, int16_t *lpc,
int16_t *buf, int16_t *dst)
{
int16_t filter_coef[2][LPC_ORDER];
@@ -655,7 +659,7 @@ static inline int cng_rand(int *state, int base)
return (*state & 0x7FFF) * base >> 15;
}
-static int estimate_sid_gain(G723_1_Context *p)
+static int estimate_sid_gain(G723_1_ChannelContext *p)
{
int i, shift, seg, seg2, t, val, val_add, x, y;
@@ -715,7 +719,7 @@ static int estimate_sid_gain(G723_1_Context *p)
return val;
}
-static void generate_noise(G723_1_Context *p)
+static void generate_noise(G723_1_ChannelContext *p)
{
int i, j, idx, t;
int off[SUBFRAMES];
@@ -843,7 +847,7 @@ static void generate_noise(G723_1_Context *p)
static int g723_1_decode_frame(AVCodecContext *avctx, void *data,
int *got_frame_ptr, AVPacket *avpkt)
{
- G723_1_Context *p = avctx->priv_data;
+ G723_1_Context *s = avctx->priv_data;
AVFrame *frame = data;
const uint8_t *buf = avpkt->data;
int buf_size = avpkt->size;
@@ -855,9 +859,8 @@ static int g723_1_decode_frame(AVCodecContext *avctx, void *data,
int16_t acb_vector[SUBFRAME_LEN];
int16_t *out;
int bad_frame = 0, i, j, ret;
- int16_t *audio = p->audio;
- if (buf_size < frame_size[dec_mode]) {
+ if (buf_size < frame_size[dec_mode] * avctx->channels) {
if (buf_size)
av_log(avctx, AV_LOG_WARNING,
"Expected %d bytes, got %d - skipping packet\n",
@@ -866,6 +869,14 @@ static int g723_1_decode_frame(AVCodecContext *avctx, void *data,
return buf_size;
}
+ frame->nb_samples = FRAME_LEN;
+ if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
+ return ret;
+
+ for (int ch = 0; ch < avctx->channels; ch++) {
+ G723_1_ChannelContext *p = &s->ch[ch];
+ int16_t *audio = p->audio;
+
if (unpack_bitstream(p, buf, buf_size) < 0) {
bad_frame = 1;
if (p->past_frame_type == ACTIVE_FRAME)
@@ -874,11 +885,7 @@ static int g723_1_decode_frame(AVCodecContext *avctx, void *data,
p->cur_frame_type = UNTRANSMITTED_FRAME;
}
- frame->nb_samples = FRAME_LEN;
- if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
- return ret;
-
- out = (int16_t *)frame->data[0];
+ out = (int16_t *)frame->extended_data[ch];
if (p->cur_frame_type == ACTIVE_FRAME) {
if (!bad_frame)
@@ -922,7 +929,7 @@ static int g723_1_decode_frame(AVCodecContext *avctx, void *data,
&p->sid_gain, &p->cur_gain);
/* Perform pitch postfiltering */
- if (p->postfilter) {
+ if (s->postfilter) {
i = PITCH_MAX;
for (j = 0; j < SUBFRAMES; i += SUBFRAME_LEN, j++)
comp_ppf_coeff(p, i, p->pitch_lag[j >> 1],
@@ -992,16 +999,17 @@ static int g723_1_decode_frame(AVCodecContext *avctx, void *data,
0, 1, 1 << 12);
memcpy(p->synth_mem, p->audio + FRAME_LEN, LPC_ORDER * sizeof(*p->audio));
- if (p->postfilter) {
+ if (s->postfilter) {
formant_postfilter(p, lpc, p->audio, out);
} else { // if output is not postfiltered it should be scaled by 2
for (i = 0; i < FRAME_LEN; i++)
out[i] = av_clip_int16(p->audio[LPC_ORDER + i] << 1);
}
+ }
*got_frame_ptr = 1;
- return frame_size[dec_mode];
+ return frame_size[dec_mode] * avctx->channels;
}
#define OFFSET(x) offsetof(G723_1_Context, x)
diff --git a/libavcodec/g723_1enc.c b/libavcodec/g723_1enc.c
index 4a4525eda9..fe3fd09c1d 100644
--- a/libavcodec/g723_1enc.c
+++ b/libavcodec/g723_1enc.c
@@ -42,7 +42,8 @@
static av_cold int g723_1_encode_init(AVCodecContext *avctx)
{
- G723_1_Context *p = avctx->priv_data;
+ G723_1_Context *s = avctx->priv_data;
+ G723_1_ChannelContext *p = &s->ch[0];
if (avctx->sample_rate != 8000) {
av_log(avctx, AV_LOG_ERROR, "Only 8000Hz sample rate supported\n");
@@ -386,7 +387,7 @@ static void iir_filter(int16_t *fir_coef, int16_t *iir_coef,
* @param flt_coef filter coefficients
* @param unq_lpc unquantized lpc vector
*/
-static void perceptual_filter(G723_1_Context *p, int16_t *flt_coef,
+static void perceptual_filter(G723_1_ChannelContext *p, int16_t *flt_coef,
int16_t *unq_lpc, int16_t *buf)
{
int16_t vector[FRAME_LEN + LPC_ORDER];
@@ -635,7 +636,7 @@ static void synth_percept_filter(int16_t *qnt_lpc, int16_t *perf_lpc,
* @param buf input signal
* @param index the current subframe index
*/
-static void acb_search(G723_1_Context *p, int16_t *residual,
+static void acb_search(G723_1_ChannelContext *p, int16_t *residual,
int16_t *impulse_resp, const int16_t *buf,
int index)
{
@@ -963,7 +964,7 @@ static void pack_fcb_param(G723_1_Subframe *subfrm, FCBParam *optim,
* @param buf target vector
* @param impulse_resp impulse response of the combined filter
*/
-static void fcb_search(G723_1_Context *p, int16_t *impulse_resp,
+static void fcb_search(G723_1_ChannelContext *p, int16_t *impulse_resp,
int16_t *buf, int index)
{
FCBParam optim;
@@ -995,7 +996,7 @@ static void fcb_search(G723_1_Context *p, int16_t *impulse_resp,
* @param frame output buffer
* @param size size of the buffer
*/
-static int pack_bitstream(G723_1_Context *p, AVPacket *avpkt)
+static int pack_bitstream(G723_1_ChannelContext *p, AVPacket *avpkt)
{
PutBitContext pb;
int info_bits = 0;
@@ -1056,7 +1057,8 @@ static int pack_bitstream(G723_1_Context *p, AVPacket *avpkt)
static int g723_1_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
const AVFrame *frame, int *got_packet_ptr)
{
- G723_1_Context *p = avctx->priv_data;
+ G723_1_Context *s = avctx->priv_data;
+ G723_1_ChannelContext *p = &s->ch[0];
int16_t unq_lpc[LPC_ORDER * SUBFRAMES];
int16_t qnt_lpc[LPC_ORDER * SUBFRAMES];
int16_t cur_lsp[LPC_ORDER];
More information about the ffmpeg-cvslog
mailing list