aacenc: add support for encoding files using Long Term Prediction

Long Term Prediction allows for prediction of spectral coefficients
via the previously decoded time-dependent samples. This feature
works well with harmonic content 2 or more frames long, like speech,
human or non-human, piano music or any constant tones at very low
bitrates.

It should be noted that the current coder is highly efficient and
the rate control system is unable to encode files at extremely
low bitrates (less than 14kbps seems to be impossible) so this
extension isn't capable of optimum operation. Dramatic difference
is observable with some types of audio and speech but for the most
part the audiable differences are subtle. The spectrum looks better
however so the encoder is able to harvest the additional bits that
this feature provies, should the user choose to enable it. So
it's best to enable this feature only if encoding at the absolutely
lowest bitrate that the encoder is capable of.

1 files changed, 43 insertions, 0 deletions

diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c
index 8d5f4387f3..9454a9491b 100644
--- a/libavcodec/aacenc.c
+++ b/libavcodec/aacenc.c
@@ -60,6 +60,7 @@ static const struct AACProfileOptions aacenc_profiles[] = {
             .mid_side = 0,
             .pns = 1,
             .tns = 0,
+            .ltp = OPT_BANNED,
             .pred = OPT_REQUIRED,
             .intensity_stereo = 1,
         },
@@ -69,6 +70,7 @@ static const struct AACProfileOptions aacenc_profiles[] = {
             .mid_side = 0,
             .pns = 1,
             .tns = 0,
+            .ltp = OPT_NEEDS_LTP,
             .pred = OPT_NEEDS_MAIN,
             .intensity_stereo = 1,
         },
@@ -78,6 +80,17 @@ static const struct AACProfileOptions aacenc_profiles[] = {
             .mid_side = 0,
             .pns = OPT_BANNED,
             .tns = 0,
+            .ltp = OPT_BANNED,
+            .pred = OPT_BANNED,
+            .intensity_stereo = 1,
+        },
+    },
+    {FF_PROFILE_AAC_LTP,
+        {  /* Long term prediction profile */
+            .mid_side = 0,
+            .pns = 1,
+            .tns = 0,
+            .ltp = OPT_REQUIRED,
             .pred = OPT_BANNED,
             .intensity_stereo = 1,
         },
@@ -475,6 +488,8 @@ static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s,
         put_ics_info(s, &sce->ics);
         if (s->coder->encode_main_pred)
             s->coder->encode_main_pred(s, sce);
+        if (s->coder->encode_ltp_info)
+            s->coder->encode_ltp_info(s, sce, 0);
     }
     encode_band_info(s, sce);
     encode_scale_factors(avctx, s, sce);
@@ -625,6 +640,13 @@ static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
             }
 
             apply_window_and_mdct(s, sce, overlap);
+
+            if (s->options.ltp && s->coder->update_ltp) {
+                s->coder->update_ltp(s, sce);
+                apply_window[sce->ics.window_sequence[0]](s->fdsp, sce, &sce->ltp_state[0]);
+                s->mdct1024.mdct_calc(&s->mdct1024, sce->lcoeffs, sce->ret_buf);
+            }
+
             if (isnan(cpe->ch->coeffs[0])) {
                 av_log(avctx, AV_LOG_ERROR, "Input contains NaN\n");
                 return AVERROR(EINVAL);
@@ -659,6 +681,8 @@ static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
                 sce = &cpe->ch[ch];
                 coeffs[ch] = sce->coeffs;
                 sce->ics.predictor_present = 0;
+                sce->ics.ltp.present = 0;
+                memset(sce->ics.ltp.used, 0, sizeof(sce->ics.ltp.used));
                 memset(sce->ics.prediction_used, 0, sizeof(sce->ics.prediction_used));
                 memset(&sce->tns, 0, sizeof(TemporalNoiseShaping));
                 for (w = 0; w < 128; w++)
@@ -738,12 +762,26 @@ static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
                 apply_mid_side_stereo(cpe);
             }
             adjust_frame_information(cpe, chans);
+            if (s->options.ltp) { /* LTP */
+                for (ch = 0; ch < chans; ch++) {
+                    sce = &cpe->ch[ch];
+                    s->cur_channel = start_ch + ch;
+                    if (s->coder->search_for_ltp)
+                        s->coder->search_for_ltp(s, sce, cpe->common_window);
+                    if (sce->ics.ltp.present) pred_mode = 1;
+                }
+                s->cur_channel = start_ch;
+                if (s->coder->adjust_common_ltp)
+                    s->coder->adjust_common_ltp(s, cpe);
+            }
             if (chans == 2) {
                 put_bits(&s->pb, 1, cpe->common_window);
                 if (cpe->common_window) {
                     put_ics_info(s, &cpe->ch[0].ics);
                     if (s->coder->encode_main_pred)
                         s->coder->encode_main_pred(s, &cpe->ch[0]);
+                    if (s->coder->encode_ltp_info)
+                        s->coder->encode_ltp_info(s, &cpe->ch[0], 1);
                     encode_ms_info(&s->pb, cpe);
                     if (cpe->ms_mode) ms_mode = 1;
                 }
@@ -816,6 +854,9 @@ static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
         }
     } while (1);
 
+    if (s->options.ltp && s->coder->ltp_insert_new_frame)
+        s->coder->ltp_insert_new_frame(s);
+
     put_bits(&s->pb, 3, TYPE_END);
     flush_put_bits(&s->pb);
     avctx->frame_bits = put_bits_count(&s->pb);
@@ -935,6 +976,7 @@ static av_cold int aac_encode_init(AVCodecContext *avctx)
     AAC_OPT_SET(&s->options, p_opt, 1, coder);
     AAC_OPT_SET(&s->options, p_opt, 0, pns);
     AAC_OPT_SET(&s->options, p_opt, 1, tns);
+    AAC_OPT_SET(&s->options, p_opt, 0, ltp);
     AAC_OPT_SET(&s->options, p_opt, 0, pred);
     AAC_OPT_SET(&s->options, p_opt, 1, mid_side);
     AAC_OPT_SET(&s->options, p_opt, 0, intensity_stereo);
@@ -993,6 +1035,7 @@ static const AVOption aacenc_options[] = {
     {"aac_is", "Intensity stereo coding", offsetof(AACEncContext, options.intensity_stereo), AV_OPT_TYPE_BOOL, {.i64 = OPT_AUTO}, -1, 1, AACENC_FLAGS},
     {"aac_pns", "Perceptual noise substitution", offsetof(AACEncContext, options.pns), AV_OPT_TYPE_BOOL, {.i64 = OPT_AUTO}, -1, 1, AACENC_FLAGS},
     {"aac_tns", "Temporal noise shaping", offsetof(AACEncContext, options.tns), AV_OPT_TYPE_BOOL, {.i64 = 0}, -1, 1, AACENC_FLAGS},
+    {"aac_ltp", "Long term prediction", offsetof(AACEncContext, options.ltp), AV_OPT_TYPE_BOOL, {.i64 = OPT_AUTO}, -1, 1, AACENC_FLAGS},
     {"aac_pred", "AAC-Main prediction", offsetof(AACEncContext, options.pred), AV_OPT_TYPE_BOOL, {.i64 = OPT_AUTO}, -1, 1, AACENC_FLAGS},
     {NULL}
 };