5  Functional description of the encoder  Word‑p. 25

5.1  Common processing

5.1.1  High-pass Filtering

5.1.2  Complex low-delay filter bank analysis

5.1.2.1  Sub-band analysis

5.1.2.2  Sub-band energy estimation  Word‑p. 26

5.1.3  Sample rate conversion to 12.8 kHz  Word‑p. 27

5.1.3.1  Conversion of 16, 32 and 48 kHz signals to 12.8 kHz

5.1.3.2  Conversion of 8 kHz signals to 12.8 kHz

5.1.3.3  Conversion of input signals to 16, 25.6 and 32 kHz  Word‑p. 29

5.1.4  Pre-emphasis

5.1.5  Spectral analysis  Word‑p. 30

5.1.5.1  Windowing and DFT

5.1.5.2  Energy calculations  Word‑p. 31

5.1.6  Bandwidth detection  Word‑p. 32

5.1.6.1  Mean and maximum energy values per band

5.1.7  Bandwidth decision  Word‑p. 34

5.1.8  Time-domain transient detection  Word‑p. 37

5.1.9  Linear prediction analysis  Word‑p. 38

5.1.9.1  LP analysis window

5.1.9.2  Autocorrelation computation

5.1.9.3  Adaptive lag windowing  Word‑p. 39

5.1.9.4  Levinson-Durbin algorithm

5.1.9.5  Conversion of LP coefficients to LSP parameters  Word‑p. 40

5.1.9.6  LSP interpolation  Word‑p. 41

5.1.9.7  Conversion of LSP parameters to LP coefficients

5.1.9.8  LP analysis at 16kHz  Word‑p. 42

5.1.10  Open-loop pitch analysis  Word‑p. 43

5.1.10.1  Perceptual weighting

5.1.10.2  Correlation function computation  Word‑p. 44

5.1.10.3  Correlation reinforcement with past pitch values  Word‑p. 45

5.1.10.4  Normalized correlation computation  Word‑p. 46

5.1.10.5  Correlation reinforcement with pitch lag multiples

5.1.10.6  Initial pitch lag determination and reinforcement based on pitch coherence with other half-frames  Word‑p. 47

5.1.10.7  Pitch lag determination and parameter update  Word‑p. 48

5.1.10.8  Correction of very short and stable open-loop pitch estimates  Word‑p. 49

5.1.10.9  Fractional open-loop pitch estimate for each subframe  Word‑p. 51

5.1.11  Background noise energy estimation  Word‑p. 52

5.1.11.1  First stage of noise energy update

5.1.11.2  Second stage of noise energy update  Word‑p. 54

5.1.11.2.1  Basic parameters for noise energy update

5.1.11.2.2  Spectral diversity  Word‑p. 55

5.1.11.2.3  Complementary non-stationarity

5.1.11.2.4  HF energy content  Word‑p. 56

5.1.11.2.5  Tonal stability

5.1.11.2.6  High frequency dynamic range  Word‑p. 60

5.1.11.2.7  Combined decision for background noise energy update

5.1.11.3  Energy-based parameters for noise energy update  Word‑p. 62

5.1.11.3.1  Closeness to current background estimate

5.1.11.3.2  Features related to last correlation or harmonic event

5.1.11.3.3  Energy-based pause detection  Word‑p. 63

5.1.11.3.4  Long-term linear prediction efficiency

5.1.11.3.5  Additional long-term parameters used for noise estimation  Word‑p. 64

5.1.11.4  Decision logic for noise energy update  Word‑p. 65

5.1.12  Signal activity detection  Word‑p. 68

5.1.12.1  SAD1 module  Word‑p. 69

5.1.12.1.1  SNR outlier filtering  Word‑p. 71

5.1.12.2  SAD2 module  Word‑p. 72

5.1.12.3  Combined decision of SAD1 and SAD2 modules for WB and SWB signals  Word‑p. 75

5.1.12.4  Final decision of the SAD1 module for NB signals

5.1.12.5  Post-decision parameter update  Word‑p. 76

5.1.12.6  SAD3 module  Word‑p. 77

5.1.12.6.1  Sub-band FFT

5.1.12.6.2  Computation of signal features  Word‑p. 78

5.1.12.6.3  Computation of SNR parameters  Word‑p. 81

5.1.12.6.4  Decision of background music  Word‑p. 83

5.1.12.6.5  Decision of background update flag

5.1.12.6.6  SAD3 Pre-decision  Word‑p. 84

5.1.12.6.7  SAD3 Hangover  Word‑p. 86

5.1.12.7  Final SAD decision

5.1.12.8  DTX hangover addition  Word‑p. 88

5.1.13  Coding mode determination  Word‑p. 90

5.1.13.1  Unvoiced signal classification  Word‑p. 91

5.1.13.1.1  Voicing measure  Word‑p. 92

5.1.13.1.2  Spectral tilt

5.1.13.1.3  Sudden energy increase from a low energy level  Word‑p. 93

5.1.13.1.4  Total frame energy difference  Word‑p. 94

5.1.13.1.5  Energy decrease after spike

5.1.13.1.6  Decision about UC mode  Word‑p. 95

5.1.13.2  Stable voiced signal classification  Word‑p. 96

5.1.13.3  Signal classification for FEC

5.1.13.3.1  Signal classes for FEC  Word‑p. 97

5.1.13.3.2  Signal classification parameters

5.1.13.3.3  Classification procedure  Word‑p. 98

5.1.13.4  Transient signal classification  Word‑p. 99

5.1.13.5  Modification of coding mode in special cases  Word‑p. 100

5.1.13.6  Speech/music classification  Word‑p. 101

5.1.13.6.1  First stage of the speech/music classifier

5.1.13.6.2  Scaling of features in the first stage of the speech/music classifier  Word‑p. 103

5.1.13.6.3  Log-probability and decision smoothing  Word‑p. 104

5.1.13.6.4  State machine and final speech/music decision  Word‑p. 105

5.1.13.6.5  Improvement of the classification for mixed and music content  Word‑p. 108

5.1.13.6.6  Second stage of the speech/music classifier  Word‑p. 112

5.1.13.6.7  Context-based improvement of the classification for stable tonal signals  Word‑p. 114

5.1.13.6.8  Detection of sparse spectral content  Word‑p. 118

5.1.13.6.9  Decision about AC mode  Word‑p. 120

5.1.13.6.10  Decision about IC mode

5.1.14  Coder technology selection

5.1.14.1  ACELP/MDCT-based technology selection at 9.6kbps, 16.4 and 24.4 kbps  Word‑p. 121

5.1.14.1.1  Segmental SNR estimation of the MDCT-based technology

5.1.14.1.2  Segmental SNR estimation of the ACELP technology  Word‑p. 127

5.1.14.1.3  Hysteresis and final decision  Word‑p. 128

5.1.14.2  TCX/HQ MDCT technology selection at 13.2 and 16.4 kbps  Word‑p. 129

5.1.14.3  TCX/HQ MDCT technology selection at 24.4 and 32 kbps  Word‑p. 131

5.1.14.4  TD/Multi-mode FD BWE technology selection at 13.2 kbps and 32 kbps  Word‑p. 134

---