wb_vad_c.h revision e2e838afcf03e603a41a0455846eaf9614537c16
1/* 2 ** Copyright 2003-2010, VisualOn, Inc. 3 ** 4 ** Licensed under the Apache License, Version 2.0 (the "License"); 5 ** you may not use this file except in compliance with the License. 6 ** You may obtain a copy of the License at 7 ** 8 ** http://www.apache.org/licenses/LICENSE-2.0 9 ** 10 ** Unless required by applicable law or agreed to in writing, software 11 ** distributed under the License is distributed on an "AS IS" BASIS, 12 ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 ** See the License for the specific language governing permissions and 14 ** limitations under the License. 15 */ 16 17 18/*-------------------------------------------------------------------* 19 * WB_VAD_C.H * 20 *-------------------------------------------------------------------* 21 * Constants for Voice Activity Detection. * 22 *-------------------------------------------------------------------*/ 23 24#ifndef __WB_VAD_C_H__ 25#define __WB_VAD_C_H__ 26 27#define FRAME_LEN 256 /* Length (samples) of the input frame */ 28#define COMPLEN 12 /* Number of sub-bands used by VAD */ 29 30#define UNIRSHFT 7 /* = log2(MAX_16/UNITY), UNITY = 256 */ 31#define SCALE 128 /* (UNITY*UNITY)/512 */ 32 33#define TONE_THR (Word16)(0.65*MAX_16) /* Threshold for tone detection */ 34 35/* constants for speech level estimation */ 36#define SP_EST_COUNT 80 37#define SP_ACTIVITY_COUNT 25 38#define ALPHA_SP_UP (Word16)((1.0 - 0.85)*MAX_16) 39#define ALPHA_SP_DOWN (Word16)((1.0 - 0.85)*MAX_16) 40 41#define NOM_LEVEL 2050 /* about -26 dBov Q15 */ 42#define SPEECH_LEVEL_INIT NOM_LEVEL /* initial speech level */ 43#define MIN_SPEECH_LEVEL1 (Word16)(NOM_LEVEL * 0.063) /* NOM_LEVEL -24 dB */ 44#define MIN_SPEECH_LEVEL2 (Word16)(NOM_LEVEL * 0.2) /* NOM_LEVEL -14 dB */ 45#define MIN_SPEECH_SNR 4096 /* 0 dB, lowest SNR estimation, Q12 */ 46 47/* Time constants for background spectrum update */ 48#define ALPHA_UP1 (Word16)((1.0 - 0.95)*MAX_16) /* Normal update, upwards: */ 49#define ALPHA_DOWN1 (Word16)((1.0 - 0.936)*MAX_16) /* Normal update, downwards */ 50#define ALPHA_UP2 (Word16)((1.0 - 0.985)*MAX_16) /* Forced update, upwards */ 51#define ALPHA_DOWN2 (Word16)((1.0 - 0.943)*MAX_16) /* Forced update, downwards */ 52#define ALPHA3 (Word16)((1.0 - 0.95)*MAX_16) /* Update downwards */ 53#define ALPHA4 (Word16)((1.0 - 0.9)*MAX_16) /* For stationary estimation */ 54#define ALPHA5 (Word16)((1.0 - 0.5)*MAX_16) /* For stationary estimation */ 55 56/* Constants for VAD threshold */ 57#define THR_MIN (Word16)(1.6*SCALE) /* Minimum threshold */ 58#define THR_HIGH (Word16)(6*SCALE) /* Highest threshold */ 59#define THR_LOW (Word16)(1.7*SCALE) /* Lowest threshold */ 60#define NO_P1 31744 /* ilog2(1), Noise level for highest threshold */ 61#define NO_P2 19786 /* ilog2(0.1*MAX_16), Noise level for lowest threshold */ 62#define NO_SLOPE (Word16)(MAX_16*(float)(THR_LOW-THR_HIGH)/(float)(NO_P2-NO_P1)) 63 64#define SP_CH_MIN (Word16)(-0.75*SCALE) 65#define SP_CH_MAX (Word16)(0.75*SCALE) 66#define SP_P1 22527 /* ilog2(NOM_LEVEL/4) */ 67#define SP_P2 17832 /* ilog2(NOM_LEVEL*4) */ 68#define SP_SLOPE (Word16)(MAX_16*(float)(SP_CH_MAX-SP_CH_MIN)/(float)(SP_P2-SP_P1)) 69 70/* Constants for hangover length */ 71#define HANG_HIGH 12 /* longest hangover */ 72#define HANG_LOW 2 /* shortest hangover */ 73#define HANG_P1 THR_LOW /* threshold for longest hangover */ 74#define HANG_P2 (Word16)(4*SCALE) /* threshold for shortest hangover */ 75#define HANG_SLOPE (Word16)(MAX_16*(float)(HANG_LOW-HANG_HIGH)/(float)(HANG_P2-HANG_P1)) 76 77/* Constants for burst length */ 78#define BURST_HIGH 8 /* longest burst length */ 79#define BURST_LOW 3 /* shortest burst length */ 80#define BURST_P1 THR_HIGH /* threshold for longest burst */ 81#define BURST_P2 THR_LOW /* threshold for shortest burst */ 82#define BURST_SLOPE (Word16)(MAX_16*(float)(BURST_LOW-BURST_HIGH)/(float)(BURST_P2-BURST_P1)) 83 84/* Parameters for background spectrum recovery function */ 85#define STAT_COUNT 20 /* threshold of stationary detection counter */ 86 87#define STAT_THR_LEVEL 184 /* Threshold level for stationarity detection */ 88#define STAT_THR 1000 /* Threshold for stationarity detection */ 89 90/* Limits for background noise estimate */ 91#define NOISE_MIN 40 /* minimum */ 92#define NOISE_MAX 20000 /* maximum */ 93#define NOISE_INIT 150 /* initial */ 94 95/* Thresholds for signal power (now calculated on 2 frames) */ 96#define VAD_POW_LOW (Word32)30000L /* If input power is lower than this, VAD is set to 0 */ 97#define POW_TONE_THR (Word32)686080L /* If input power is lower,tone detection flag is ignored */ 98 99/* Constants for the filter bank */ 100#define COEFF3 13363 /* coefficient for the 3rd order filter */ 101#define COEFF5_1 21955 /* 1st coefficient the for 5th order filter */ 102#define COEFF5_2 6390 /* 2nd coefficient the for 5th order filter */ 103#define F_5TH_CNT 5 /* number of 5th order filters */ 104#define F_3TH_CNT 6 /* number of 3th order filters */ 105 106#endif //__WB_VAD_C_H__ 107 108 109 110