[FFmpeg-devel] [PATCH 1/2][RFC] avcodec/g729dec: move definitions to header file
Ganesh Ajjanagadde
gajjanagadde at gmail.com
Tue Aug 11 03:54:14 CEST 2015
On Mon, Aug 10, 2015 at 9:51 PM, Ganesh Ajjanagadde
<gajjanagadde at gmail.com> wrote:
> Moves structure definitions and related macros to g729dec.h
> Also exports format in priv_data
>
> Signed-off-by: Ganesh Ajjanagadde <gajjanagadde at gmail.com>
> ---
> libavcodec/g729dec.c | 111 ++---------------------------------------
> libavcodec/g729dec.h | 138 +++++++++++++++++++++++++++++++++++++++++++++++++++
> 2 files changed, 141 insertions(+), 108 deletions(-)
> create mode 100644 libavcodec/g729dec.h
>
> diff --git a/libavcodec/g729dec.c b/libavcodec/g729dec.c
> index 99053ad..e5b2de0 100644
> --- a/libavcodec/g729dec.c
> +++ b/libavcodec/g729dec.c
> @@ -30,6 +30,7 @@
>
>
> #include "g729.h"
> +#include "g729dec.h"
> #include "lsp.h"
> #include "celp_math.h"
> #include "celp_filters.h"
> @@ -39,57 +40,6 @@
> #include "g729data.h"
> #include "g729postfilter.h"
>
> -/**
> - * minimum quantized LSF value (3.2.4)
> - * 0.005 in Q13
> - */
> -#define LSFQ_MIN 40
> -
> -/**
> - * maximum quantized LSF value (3.2.4)
> - * 3.135 in Q13
> - */
> -#define LSFQ_MAX 25681
> -
> -/**
> - * minimum LSF distance (3.2.4)
> - * 0.0391 in Q13
> - */
> -#define LSFQ_DIFF_MIN 321
> -
> -/// interpolation filter length
> -#define INTERPOL_LEN 11
> -
> -/**
> - * minimum gain pitch value (3.8, Equation 47)
> - * 0.2 in (1.14)
> - */
> -#define SHARP_MIN 3277
> -
> -/**
> - * maximum gain pitch value (3.8, Equation 47)
> - * (EE) This does not comply with the specification.
> - * Specification says about 0.8, which should be
> - * 13107 in (1.14), but reference C code uses
> - * 13017 (equals to 0.7945) instead of it.
> - */
> -#define SHARP_MAX 13017
> -
> -/**
> - * MR_ENERGY (mean removed energy) = mean_energy + 10 * log10(2^26 * subframe_size) in (7.13)
> - */
> -#define MR_ENERGY 1018156
> -
> -#define DECISION_NOISE 0
> -#define DECISION_INTERMEDIATE 1
> -#define DECISION_VOICE 2
> -
> -typedef enum {
> - FORMAT_G729_8K = 0,
> - FORMAT_G729D_6K4,
> - FORMAT_COUNT,
> -} G729Formats;
> -
> typedef struct {
> uint8_t ac_index_bits[2]; ///< adaptive codebook index for second subframe (size in bits)
> uint8_t parity_bit; ///< parity bit for pitch delay
> @@ -99,61 +49,6 @@ typedef struct {
> uint8_t fc_indexes_bits; ///< size (in bits) of fixed-codebook index entry
> } G729FormatDescription;
>
> -typedef struct {
> - AudioDSPContext adsp;
> -
> - /// past excitation signal buffer
> - int16_t exc_base[2*SUBFRAME_SIZE+PITCH_DELAY_MAX+INTERPOL_LEN];
> -
> - int16_t* exc; ///< start of past excitation data in buffer
> - int pitch_delay_int_prev; ///< integer part of previous subframe's pitch delay (4.1.3)
> -
> - /// (2.13) LSP quantizer outputs
> - int16_t past_quantizer_output_buf[MA_NP + 1][10];
> - int16_t* past_quantizer_outputs[MA_NP + 1];
> -
> - int16_t lsfq[10]; ///< (2.13) quantized LSF coefficients from previous frame
> - int16_t lsp_buf[2][10]; ///< (0.15) LSP coefficients (previous and current frames) (3.2.5)
> - int16_t *lsp[2]; ///< pointers to lsp_buf
> -
> - int16_t quant_energy[4]; ///< (5.10) past quantized energy
> -
> - /// previous speech data for LP synthesis filter
> - int16_t syn_filter_data[10];
> -
> -
> - /// residual signal buffer (used in long-term postfilter)
> - int16_t residual[SUBFRAME_SIZE + RES_PREV_DATA_SIZE];
> -
> - /// previous speech data for residual calculation filter
> - int16_t res_filter_data[SUBFRAME_SIZE+10];
> -
> - /// previous speech data for short-term postfilter
> - int16_t pos_filter_data[SUBFRAME_SIZE+10];
> -
> - /// (1.14) pitch gain of current and five previous subframes
> - int16_t past_gain_pitch[6];
> -
> - /// (14.1) gain code from current and previous subframe
> - int16_t past_gain_code[2];
> -
> - /// voice decision on previous subframe (0-noise, 1-intermediate, 2-voice), G.729D
> - int16_t voice_decision;
> -
> - int16_t onset; ///< detected onset level (0-2)
> - int16_t was_periodic; ///< whether previous frame was declared as periodic or not (4.4)
> - int16_t ht_prev_data; ///< previous data for 4.2.3, equation 86
> - int gain_coeff; ///< (1.14) gain coefficient (4.2.4)
> - uint16_t rand_value; ///< random number generator value (4.4.4)
> - int ma_predictor_prev; ///< switched MA predictor of LSP quantizer from last good frame
> -
> - /// (14.14) high-pass filter data (past input)
> - int hpf_f[2];
> -
> - /// high-pass filter data (past output)
> - int16_t hpf_z[2];
> -} G729Context;
> -
> static const G729FormatDescription format_g729_8k = {
> .ac_index_bits = {8,5},
> .parity_bit = 1,
> @@ -422,14 +317,14 @@ static int decode_frame(AVCodecContext *avctx, void *data, int *got_frame_ptr,
> out_frame = (int16_t*) frame->data[0];
>
> if (buf_size % 10 == 0) {
> - packet_type = FORMAT_G729_8K;
> + ctx->format = packet_type = FORMAT_G729_8K;
> format = &format_g729_8k;
> //Reset voice decision
> ctx->onset = 0;
> ctx->voice_decision = DECISION_VOICE;
> av_log(avctx, AV_LOG_DEBUG, "Packet type: %s\n", "G.729 @ 8kbit/s");
> } else if (buf_size == 8) {
> - packet_type = FORMAT_G729D_6K4;
> + ctx->format = packet_type = FORMAT_G729D_6K4;
> format = &format_g729d_6k4;
> av_log(avctx, AV_LOG_DEBUG, "Packet type: %s\n", "G.729D @ 6.4kbit/s");
> } else {
> diff --git a/libavcodec/g729dec.h b/libavcodec/g729dec.h
> new file mode 100644
> index 0000000..06cbebf
> --- /dev/null
> +++ b/libavcodec/g729dec.h
> @@ -0,0 +1,138 @@
> +/*
> + * G.729 decoder
> + * Copyright (C) 2015 Ganesh Ajjanagadde
> + *
> + * This file is part of FFmpeg.
> + *
> + * FFmpeg is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * FFmpeg 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
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with FFmpeg; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
> + */
> +
> +#ifndef AVCODEC_G729DEC_H
> +#define AVCODEC_G729DEC_H
> +
> +#include "acelp_pitch_delay.h"
> +#include "g729data.h"
> +#include "g729postfilter.h"
> +
> +typedef enum {
> + FORMAT_G729_8K = 0,
> + FORMAT_G729D_6K4,
> + FORMAT_COUNT,
> +} G729Formats;
> +
> +/**
> + * minimum quantized LSF value (3.2.4)
> + * 0.005 in Q13
> + */
> +#define LSFQ_MIN 40
> +
> +/**
> + * maximum quantized LSF value (3.2.4)
> + * 3.135 in Q13
> + */
> +#define LSFQ_MAX 25681
> +
> +/**
> + * minimum LSF distance (3.2.4)
> + * 0.0391 in Q13
> + */
> +#define LSFQ_DIFF_MIN 321
> +
> +/// interpolation filter length
> +#define INTERPOL_LEN 11
> +
> +/**
> + * minimum gain pitch value (3.8, Equation 47)
> + * 0.2 in (1.14)
> + */
> +#define SHARP_MIN 3277
> +
> +/**
> + * maximum gain pitch value (3.8, Equation 47)
> + * (EE) This does not comply with the specification.
> + * Specification says about 0.8, which should be
> + * 13107 in (1.14), but reference C code uses
> + * 13017 (equals to 0.7945) instead of it.
> + */
> +#define SHARP_MAX 13017
> +
> +/**
> + * MR_ENERGY (mean removed energy) = mean_energy + 10 * log10(2^26 * subframe_size) in (7.13)
> + */
> +#define MR_ENERGY 1018156
> +
> +#define DECISION_NOISE 0
> +#define DECISION_INTERMEDIATE 1
> +#define DECISION_VOICE 2
> +
> +typedef struct {
> + AudioDSPContext adsp;
> +
> + /// past excitation signal buffer
> + int16_t exc_base[2*SUBFRAME_SIZE+PITCH_DELAY_MAX+INTERPOL_LEN];
> +
> + int16_t* exc; ///< start of past excitation data in buffer
> + int pitch_delay_int_prev; ///< integer part of previous subframe's pitch delay (4.1.3)
> +
> + /// (2.13) LSP quantizer outputs
> + int16_t past_quantizer_output_buf[MA_NP + 1][10];
> + int16_t* past_quantizer_outputs[MA_NP + 1];
> +
> + int16_t lsfq[10]; ///< (2.13) quantized LSF coefficients from previous frame
> + int16_t lsp_buf[2][10]; ///< (0.15) LSP coefficients (previous and current frames) (3.2.5)
> + int16_t *lsp[2]; ///< pointers to lsp_buf
> +
> + int16_t quant_energy[4]; ///< (5.10) past quantized energy
> +
> + /// previous speech data for LP synthesis filter
> + int16_t syn_filter_data[10];
> +
> +
> + /// residual signal buffer (used in long-term postfilter)
> + int16_t residual[SUBFRAME_SIZE + RES_PREV_DATA_SIZE];
> +
> + /// previous speech data for residual calculation filter
> + int16_t res_filter_data[SUBFRAME_SIZE+10];
> +
> + /// previous speech data for short-term postfilter
> + int16_t pos_filter_data[SUBFRAME_SIZE+10];
> +
> + /// (1.14) pitch gain of current and five previous subframes
> + int16_t past_gain_pitch[6];
> +
> + /// (14.1) gain code from current and previous subframe
> + int16_t past_gain_code[2];
> +
> + /// voice decision on previous subframe (0-noise, 1-intermediate, 2-voice), G.729D
> + int16_t voice_decision;
> +
> + int16_t onset; ///< detected onset level (0-2)
> + int16_t was_periodic; ///< whether previous frame was declared as periodic or not (4.4)
> + int16_t ht_prev_data; ///< previous data for 4.2.3, equation 86
> + int gain_coeff; ///< (1.14) gain coefficient (4.2.4)
> + uint16_t rand_value; ///< random number generator value (4.4.4)
> + int ma_predictor_prev; ///< switched MA predictor of LSP quantizer from last good frame
> +
> + /// (14.14) high-pass filter data (past input)
> + int hpf_f[2];
> +
> + /// high-pass filter data (past output)
> + int16_t hpf_z[2];
> +
> + /// format type
> + G729Formats format;
> +} G729Context;
> +
> +#endif /* AVCODEC_G729DEC_H */
> --
> 2.5.0
>
Please note this is just a rough patch that together with the second
fixes Ticket4753.
I am not familiar with how to keep things private, in scope and avoid
unnecessary exposure of api, etc,
and would greatly appreciate help especially in this aspect.
More information about the ffmpeg-devel
mailing list