cnst_vad.h

Trolltech公司发布的图形界面操作系统。可在qt-embedded-2.3.10平台上编译为嵌入式图形界面操作系统。

/**********************************************************************
** Copyright (C) 2000-2005 Trolltech AS and its licensors.
** All rights reserved.
**
** This file is part of the Qtopia Environment.
**
** This file may be distributed and/or modified under the terms of the
** GNU General Public License version 2 as published by the Free Software
** Foundation and appearing in the file LICENSE.GPL included in the
** packaging of this file.
**
** This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
** WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
**
** See http://www.trolltech.com/gpl/ for GPL licensing information.
** See below for additional copyright and license information
**
** Contact [e-mail] if any conditions of this licensing are
** not clear to you.
**
**********************************************************************/
/*
********************************************************************************
**-------------------------------------------------------------------------**
**                                                                         **
**     GSM AMR-NB speech codec   R98   Version 7.6.0   December 12, 2001       **
**                               R99   Version 3.3.0                       **
**                               REL-4 Version 4.1.0                       **
**                                                                         **
**-------------------------------------------------------------------------**
********************************************************************************
*
*      File             : cnst_vad.h
*      Purpose          : Constants and definitions for VAD
*
********************************************************************************
*/
#ifndef cnst_vad_h
#define cnst_vad_h "$Id $"

#define FRAME_LEN 160    /* Length (samples) of the input frame          */
#define COMPLEN 9        /* Number of sub-bands used by VAD              */
#define INV_COMPLEN 3641 /* 1.0/COMPLEN*2^15                             */
#define LOOKAHEAD 40     /* length of the lookahead used by speech coder */

#define UNITY 512        /* Scaling used with SNR calculation            */
#define UNIRSHFT 6       /* = log2(MAX_16/UNITY)                         */

#define TONE_THR (Word16)(0.65*MAX_16) /* Threshold for tone detection   */

/* Constants for background spectrum update */
#define ALPHA_UP1   (Word16)((1.0 - 0.95)*MAX_16)  /* Normal update, upwards:   */
#define ALPHA_DOWN1 (Word16)((1.0 - 0.936)*MAX_16) /* Normal update, downwards  */
#define ALPHA_UP2   (Word16)((1.0 - 0.985)*MAX_16) /* Forced update, upwards    */
#define ALPHA_DOWN2 (Word16)((1.0 - 0.943)*MAX_16) /* Forced update, downwards  */
#define ALPHA3      (Word16)((1.0 - 0.95)*MAX_16)  /* Update downwards          */
#define ALPHA4      (Word16)((1.0 - 0.9)*MAX_16)   /* For stationary estimation */
#define ALPHA5      (Word16)((1.0 - 0.5)*MAX_16)   /* For stationary estimation */

/* Constants for VAD threshold */
#define VAD_THR_HIGH 1260 /* Highest threshold                 */
#define VAD_THR_LOW  720  /* Lowest threshold                  */
#define VAD_P1 0          /* Noise level for highest threshold */
#define VAD_P2 6300       /* Noise level for lowest threshold  */
#define VAD_SLOPE (Word16)(MAX_16*(float)(VAD_THR_LOW-VAD_THR_HIGH)/(float)(VAD_P2-VAD_P1))

/* Parameters for background spectrum recovery function */
#define STAT_COUNT 20         /* threshold of stationary detection counter         */
#define STAT_COUNT_BY_2 10    /* threshold of stationary detection counter         */
#define CAD_MIN_STAT_COUNT 5  /* threshold of stationary detection counter         */

#define STAT_THR_LEVEL 184    /* Threshold level for stationarity detection        */
#define STAT_THR 1000         /* Threshold for stationarity detection              */

/* Limits for background noise estimate */
#define NOISE_MIN 40          /* minimum */
#define NOISE_MAX 16000       /* maximum */
#define NOISE_INIT 150        /* initial */

/* Constants for VAD hangover addition */
#define HANG_NOISE_THR 100
#define BURST_LEN_HIGH_NOISE 4
#define HANG_LEN_HIGH_NOISE 7
#define BURST_LEN_LOW_NOISE 5
#define HANG_LEN_LOW_NOISE 4

/* Thresholds for signal power */
#define VAD_POW_LOW (Word32)15000     /* If input power is lower,                    */
                                      /*     VAD is set to 0                         */
#define POW_PITCH_THR (Word32)343040  /* If input power is lower, pitch              */
                                      /*     detection is ignored                    */

#define POW_COMPLEX_THR (Word32)15000 /* If input power is lower, complex            */
                                      /* flags  value for previous frame  is un-set  */
 

/* Constants for the filter bank */
#define LEVEL_SHIFT 0      /* scaling                                  */
#define COEFF3   13363     /* coefficient for the 3rd order filter     */
#define COEFF5_1 21955     /* 1st coefficient the for 5th order filter */
#define COEFF5_2 6390      /* 2nd coefficient the for 5th order filter */

/* Constants for pitch detection */
#define LTHRESH 4
#define NTHRESH 4

/* Constants for complex signal VAD  */
#define CVAD_THRESH_ADAPT_HIGH  (Word16)(0.6 * MAX_16) /* threshold for adapt stopping high    */
#define CVAD_THRESH_ADAPT_LOW  (Word16)(0.5 * MAX_16)  /* threshold for adapt stopping low     */
#define CVAD_THRESH_IN_NOISE  (Word16)(0.65 * MAX_16)  /* threshold going into speech on
                                                          a short term basis                   */

#define CVAD_THRESH_HANG  (Word16)(0.70 * MAX_16)      /* threshold                            */
#define CVAD_HANG_LIMIT  (Word16)(100)                 /* 2 second estimation time             */
#define CVAD_HANG_LENGTH  (Word16)(250)                /* 5 second hangover                    */

#define CVAD_LOWPOW_RESET (Word16) (0.40 * MAX_16)     /* init in low power segment            */
#define CVAD_MIN_CORR (Word16) (0.40 * MAX_16)         /* lowest adaptation value              */

#define CVAD_BURST 20                                  /* speech burst length for speech reset */
#define CVAD_ADAPT_SLOW (Word16)(( 1.0 - 0.98) * MAX_16)        /* threshold for slow adaption */
#define CVAD_ADAPT_FAST (Word16)((1.0 - 0.92) * MAX_16)         /* threshold for fast adaption */
#define CVAD_ADAPT_REALLY_FAST (Word16)((1.0 - 0.80) * MAX_16)  /* threshold for really fast
                                                                   adaption                    */

#endif