FreeCalypso > hg > gsmhr-codec-ref
view sp_enc.c @ 6:9cbb19619a9f default tip
README: punctuation fix
author | Mychaela Falconia <falcon@freecalypso.org> |
---|---|
date | Tue, 20 Aug 2024 19:00:23 +0000 |
parents | 9008dbc8ca74 |
children |
line wrap: on
line source
/*************************************************************************** * * File Name: sp_enc.c * * Purpose: Contains speech encoder function. Calls are made to the * frame-based encoding functions (see sp_frm.c), and the subframe- * based encoding function (see sp_sfrm.c) * * Functions in this file (only 1) * speechEncoder() * **************************************************************************/ /*_________________________________________________________________________ | | | Include Files | |_________________________________________________________________________| */ #include "mathhalf.h" #include "mathdp31.h" #include "sp_rom.h" #include "sp_dec.h" #include "sp_frm.h" #include "sp_sfrm.h" #include "sp_enc.h" #include "host.h" #include "vad.h" /*_________________________________________________________________________ | | | Local Defines | |_________________________________________________________________________| */ #define CG_INT_MACS 6 /* Number of Multiply-Accumulates in */ /* one interpolation */ #define ASCALE 0x0800 #define LMAX 142 /* largest lag (integer sense) */ #define LSMAX (LMAX+ CG_INT_MACS/2) /* Lag Search Array Length */ #define NUM_CLOSED 3 /* Maximum number of lags searched */ /* in closed loop. */ #define LPCSTARTINDEX 25 /* Where the LPC analysis window * starts */ #define INBUFFSZ LPCSTARTINDEX + A_LEN /* Input buffer size */ #define NUMSTARTUPSMP INBUFFSZ - F_LEN /* Number of samples needed */ /* at start up */ #define NUMSTARTUPSMP_P1 INBUFFSZ - F_LEN + 1 #define HPFSHIFT 1 /* no right shifts high pass shifts * speech */ /*_________________________________________________________________________ | | | State variables (globals) | |_________________________________________________________________________| */ Shortword swOldR0; Shortword swOldR0Index; struct NormSw psnsWSfrmEngSpace[2 * N_SUB]; Shortword pswHPFXState[4]; Shortword pswHPFYState[8]; Shortword pswOldFrmKs[NP]; Shortword pswOldFrmAs[NP]; Shortword pswOldFrmSNWCoefs[NP]; Shortword pswWgtSpeechSpace[F_LEN + LMAX + CG_INT_MACS / 2]; Shortword pswSpeech[INBUFFSZ]; /* input speech */ Shortword swPtch = 1; /*_________________________________________________________________________ | | | Global DTX variables | |_________________________________________________________________________| */ Shortword swTxGsHistPtr = 0; Shortword pswCNVSCode1[N_SUB], pswCNVSCode2[N_SUB], pswCNGsp0Code[N_SUB], pswCNLpc[3], swCNR0; extern Longword pL_GsHist[]; extern LongwordRom ppLr_gsTable[4][32]; /*************************************************************************** * * FUNCTION NAME: speechEncoder * * PURPOSE: * * Performs GSM half-rate speech encoding on frame basis (160 samples). * * INPUTS: * * pswSpeechIn[0:159] - input speech samples, 160 new samples per frame * * OUTPUTS: * * pswFrmCodes[0:19] - output parameters, 18 speech parameters plus * VAD and SP flags * * RETURN VALUE: * * None * * IMPLEMENTATION: * * n/a * * REFERENCES: Sub-clause 4.1 of GSM Recomendation 06.20 * * KEYWORDS: speechcoder, analysis * *************************************************************************/ void speechEncoder(Shortword pswSpeechIn[], Shortword pswFrmCodes[]) { /*_________________________________________________________________________ | | | Static Variables | |_________________________________________________________________________| */ /* 1st point in analysis window */ static Shortword *pswLpcStart = &pswSpeech[LPCSTARTINDEX]; /* 1st point of new frame other than 1st */ static Shortword *pswNewSpeech = &pswSpeech[NUMSTARTUPSMP]; /* sample 0 of weighted speech */ static Shortword *pswWgtSpeech = &pswWgtSpeechSpace[LSMAX]; static struct NormSw *psnsWSfrmEng = &psnsWSfrmEngSpace[N_SUB]; /*_________________________________________________________________________ | | | Automatic Variables | |_________________________________________________________________________| */ int iVoicing, /* bitAlloc */ iR0, /* bitAlloc and aflat */ piVq[3], /* bitAlloc */ iSi, /* bitAlloc */ piLagCode[N_SUB], /* bitAlloc */ piVSCode1[N_SUB], /* bitAlloc */ piVSCode2[N_SUB], /* bitAlloc */ piGsp0Code[N_SUB]; /* bitAlloc */ short int siUVCode, siSi, i, j; Shortword swR0, pswLagCode[N_SUB], pswVSCode1[N_SUB], pswVSCode2[N_SUB], pswGsp0Code[N_SUB], *pswLagListPtr, pswFrmKs[NP], pswFrmAs[NP], pswFrmSNWCoefs[NP], pswLagList[N_SUB * NUM_CLOSED], pswNumLagList[N_SUB], pswPitchBuf[N_SUB], pswHNWCoefBuf[N_SUB], ppswSNWCoefAs[N_SUB][NP], ppswSynthAs[N_SUB][NP]; Shortword swSP, pswVadLags[4], /* VAD Parameters */ swVadFlag; /* flag indicating voice activity * detector state. 1 = speech or * speech/signal present */ struct NormSw psnsSqrtRs[N_SUB]; /*_________________________________________________________________________ | | | Executable Code | |_________________________________________________________________________| */ /* Speech frame processing */ /* High pass filter the speech */ /* ---------------------------- */ filt4_2nd(psrHPFCoefs, pswSpeechIn, pswHPFXState, pswHPFYState, F_LEN, HPFSHIFT); /* copy high passed filtered speech into encoder's speech buff */ /*-------------------------------------------------------------*/ for (i = 0; i < F_LEN; i++) pswNewSpeech[i] = pswSpeechIn[i]; /* Calculate and quantize LPC coefficients */ /* --------------------------------------- */ aflat(pswLpcStart, &iR0, pswFrmKs, piVq, swPtch, &swVadFlag, &swSP); /* Lookup frame energy r0 */ /* ---------------------- */ swR0 = psrR0DecTbl[iR0 * 2]; /* lookupR0 */ /* Generate the direct form coefs */ /* ------------------------------ */ if (!rcToADp(ASCALE, pswFrmKs, pswFrmAs)) { getNWCoefs(pswFrmAs, pswFrmSNWCoefs); } else { for (i = 0; i < NP; i++) { pswFrmKs[i] = pswOldFrmKs[i]; pswFrmAs[i] = pswOldFrmAs[i]; pswFrmSNWCoefs[i] = pswOldFrmSNWCoefs[i]; } } /* Interpolate, or otherwise get sfrm reflection coefs */ /* --------------------------------------------------- */ getSfrmLpcTx(swOldR0, swR0, pswOldFrmKs, pswOldFrmAs, pswOldFrmSNWCoefs, pswFrmKs, pswFrmAs, pswFrmSNWCoefs, pswSpeech, &siSi, psnsSqrtRs, ppswSynthAs, ppswSNWCoefAs); /* loose once bitAlloc done */ iSi = siSi; /* Weight the entire speech frame */ /* ------------------------------ */ weightSpeechFrame(pswSpeech, ppswSynthAs[0], ppswSNWCoefAs[0], pswWgtSpeechSpace); /* Perform open-loop lag search, get harmonic-noise-weighting parameters */ /* --------------------------------------------------------------------- */ openLoopLagSearch(&pswWgtSpeechSpace[LSMAX], swOldR0Index, (Shortword) iR0, &siUVCode, pswLagList, pswNumLagList, pswPitchBuf, pswHNWCoefBuf, &psnsWSfrmEng[0], pswVadLags, swSP); iVoicing = siUVCode; /* Using open loop LTP data to calculate swPtch */ /* DTX mode */ /* parameter */ /* DTX mode */ /* -------------------------------------------- */ /* DTX mode */ periodicity_update(pswVadLags, &swPtch); /* DTX mode */ /* Subframe processing loop */ /* ------------------------ */ pswLagListPtr = pswLagList; for (giSfrmCnt = 0; giSfrmCnt < N_SUB; giSfrmCnt++) { if (swSP == 0) /* DTX mode */ { /* DTX mode */ pswVSCode1[giSfrmCnt] = pswCNVSCode1[giSfrmCnt]; /* DTX mode */ pswVSCode2[giSfrmCnt] = pswCNVSCode2[giSfrmCnt]; /* DTX mode */ pswGsp0Code[giSfrmCnt] = pswCNGsp0Code[giSfrmCnt]; /* DTX mode */ } /* DTX mode */ sfrmAnalysis(&pswWgtSpeech[giSfrmCnt * S_LEN], siUVCode, psnsSqrtRs[giSfrmCnt], ppswSNWCoefAs[giSfrmCnt], pswLagListPtr, pswNumLagList[giSfrmCnt], pswPitchBuf[giSfrmCnt], pswHNWCoefBuf[giSfrmCnt], &pswLagCode[giSfrmCnt], &pswVSCode1[giSfrmCnt], &pswVSCode2[giSfrmCnt], &pswGsp0Code[giSfrmCnt], swSP); pswLagListPtr = &pswLagListPtr[pswNumLagList[giSfrmCnt]]; } /* copy comfort noise parameters, */ /* DTX mode */ /* update GS history */ /* DTX mode */ /* ------------------------------ */ /* DTX mode */ if (swSP == 0) /* DTX mode */ { /* DTX mode */ /* copy comfort noise frame parameter */ /* DTX mode */ /* ---------------------------------- */ /* DTX mode */ iR0 = swCNR0; /* quantized R0 index */ /* DTX mode */ for (i=0; i < 3; i++) /* DTX mode */ piVq[i] = pswCNLpc[i]; /* DTX mode */ } /* DTX mode */ else /* DTX mode */ { /* DTX mode */ /* if swSP != 0, then update the GS history */ /* DTX mode */ /* -----------------------------------------*/ /* DTX mode */ for (i=0; i < N_SUB; i++){ /* DTX mode */ pL_GsHist[swTxGsHistPtr] = /* DTX mode */ ppLr_gsTable[siUVCode][pswGsp0Code[i]]; /* DTX mode */ swTxGsHistPtr++; /* DTX mode */ if (swTxGsHistPtr > ((OVERHANG-1)*N_SUB)-1) /* DTX mode */ swTxGsHistPtr=0; /* DTX mode */ } /* DTX mode */ } /* DTX mode */ /* End of frame processing, update frame based parameters */ /* ------------------------------------------------------ */ for (i = 0; i < N_SUB; i++) { piLagCode[i] = pswLagCode[i]; piVSCode1[i] = pswVSCode1[i]; piVSCode2[i] = pswVSCode2[i]; piGsp0Code[i] = pswGsp0Code[i]; } swOldR0Index = (Shortword) iR0; swOldR0 = swR0; for (i = 0; i < NP; i++) { pswOldFrmKs[i] = pswFrmKs[i]; pswOldFrmAs[i] = pswFrmAs[i]; pswOldFrmSNWCoefs[i] = pswFrmSNWCoefs[i]; } /* Insert SID Codeword */ /* DTX mode */ /* ------------------- */ /* DTX mode */ if (swSP == 0) /* DTX mode */ { /* DTX mode */ iVoicing = 0x0003; /* 2 bits */ /* DTX mode */ iSi = 0x0001; /* 1 bit */ /* DTX mode */ for (i=0; i < N_SUB; i++) /* DTX mode */ { /* DTX mode */ piVSCode1[i] = 0x01ff; /* 9 bits */ /* DTX mode */ piGsp0Code[i] = 0x001f; /* 5 bits */ /* DTX mode */ } piLagCode[0] = 0x00ff; /* 8 bits */ /* DTX mode */ piLagCode[1] = 0x000f; /* 4 bits */ /* DTX mode */ piLagCode[2] = 0x000f; /* 4 bits */ /* DTX mode */ piLagCode[3] = 0x000f; /* 4 bits */ /* DTX mode */ } /* DTX mode */ /* Generate encoded parameter array */ /* -------------------------------- */ fillBitAlloc(iVoicing, iR0, piVq, iSi, piLagCode, piVSCode1, piVSCode2, piGsp0Code, swVadFlag, swSP, pswFrmCodes); /* delay the input speech by 1 frame */ /*-----------------------------------*/ for (i = 0, j = F_LEN; j < INBUFFSZ; i++, j++) { pswSpeech[i] = pswSpeech[j]; } }