blob: a76ed7d9fe2b9eb98f07679a8d449f1a655f2632 [file] [log] [blame]
/*
* Copyright (c) 2011 The WebRTC project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "webrtc/modules/audio_coding/codecs/isac/fix/source/pitch_estimator.h"
#include "webrtc/common_audio/signal_processing/include/signal_processing_library.h"
#include "webrtc/rtc_base/compile_assert_c.h"
extern int32_t WebRtcIsacfix_Log2Q8(uint32_t x);
void WebRtcIsacfix_PCorr2Q32(const int16_t* in, int32_t* logcorQ8) {
int16_t scaling,n,k;
int32_t ysum32,csum32, lys, lcs;
const int32_t oneQ8 = 1 << 8; // 1.00 in Q8
const int16_t* x;
const int16_t* inptr;
x = in + PITCH_MAX_LAG / 2 + 2;
scaling = WebRtcSpl_GetScalingSquare((int16_t*)in,
PITCH_CORR_LEN2,
PITCH_CORR_LEN2);
ysum32 = 1;
csum32 = 0;
x = in + PITCH_MAX_LAG / 2 + 2;
{
const int16_t* tmp_x = x;
const int16_t* tmp_in = in;
int32_t tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8;
n = PITCH_CORR_LEN2;
RTC_COMPILE_ASSERT(PITCH_CORR_LEN2 % 4 == 0);
__asm __volatile (
".set push \n\t"
".set noreorder \n\t"
"1: \n\t"
"lh %[tmp1], 0(%[tmp_in]) \n\t"
"lh %[tmp2], 2(%[tmp_in]) \n\t"
"lh %[tmp3], 4(%[tmp_in]) \n\t"
"lh %[tmp4], 6(%[tmp_in]) \n\t"
"lh %[tmp5], 0(%[tmp_x]) \n\t"
"lh %[tmp6], 2(%[tmp_x]) \n\t"
"lh %[tmp7], 4(%[tmp_x]) \n\t"
"lh %[tmp8], 6(%[tmp_x]) \n\t"
"mul %[tmp5], %[tmp1], %[tmp5] \n\t"
"mul %[tmp1], %[tmp1], %[tmp1] \n\t"
"mul %[tmp6], %[tmp2], %[tmp6] \n\t"
"mul %[tmp2], %[tmp2], %[tmp2] \n\t"
"mul %[tmp7], %[tmp3], %[tmp7] \n\t"
"mul %[tmp3], %[tmp3], %[tmp3] \n\t"
"mul %[tmp8], %[tmp4], %[tmp8] \n\t"
"mul %[tmp4], %[tmp4], %[tmp4] \n\t"
"addiu %[n], %[n], -4 \n\t"
"srav %[tmp5], %[tmp5], %[scaling] \n\t"
"srav %[tmp1], %[tmp1], %[scaling] \n\t"
"srav %[tmp6], %[tmp6], %[scaling] \n\t"
"srav %[tmp2], %[tmp2], %[scaling] \n\t"
"srav %[tmp7], %[tmp7], %[scaling] \n\t"
"srav %[tmp3], %[tmp3], %[scaling] \n\t"
"srav %[tmp8], %[tmp8], %[scaling] \n\t"
"srav %[tmp4], %[tmp4], %[scaling] \n\t"
"addu %[ysum32], %[ysum32], %[tmp1] \n\t"
"addu %[csum32], %[csum32], %[tmp5] \n\t"
"addu %[ysum32], %[ysum32], %[tmp2] \n\t"
"addu %[csum32], %[csum32], %[tmp6] \n\t"
"addu %[ysum32], %[ysum32], %[tmp3] \n\t"
"addu %[csum32], %[csum32], %[tmp7] \n\t"
"addu %[ysum32], %[ysum32], %[tmp4] \n\t"
"addu %[csum32], %[csum32], %[tmp8] \n\t"
"addiu %[tmp_in], %[tmp_in], 8 \n\t"
"bgtz %[n], 1b \n\t"
" addiu %[tmp_x], %[tmp_x], 8 \n\t"
".set pop \n\t"
: [tmp1] "=&r" (tmp1), [tmp2] "=&r" (tmp2), [tmp3] "=&r" (tmp3),
[tmp4] "=&r" (tmp4), [tmp5] "=&r" (tmp5), [tmp6] "=&r" (tmp6),
[tmp7] "=&r" (tmp7), [tmp8] "=&r" (tmp8), [tmp_in] "+r" (tmp_in),
[ysum32] "+r" (ysum32), [tmp_x] "+r" (tmp_x), [csum32] "+r" (csum32),
[n] "+r" (n)
: [scaling] "r" (scaling)
: "memory", "hi", "lo"
);
}
logcorQ8 += PITCH_LAG_SPAN2 - 1;
lys = WebRtcIsacfix_Log2Q8((uint32_t)ysum32) >> 1; // Q8, sqrt(ysum)
if (csum32 > 0) {
lcs = WebRtcIsacfix_Log2Q8((uint32_t)csum32); // 2log(csum) in Q8
if (lcs > (lys + oneQ8)) { // csum/sqrt(ysum) > 2 in Q8
*logcorQ8 = lcs - lys; // log2(csum/sqrt(ysum))
} else {
*logcorQ8 = oneQ8; // 1.00
}
} else {
*logcorQ8 = 0;
}
for (k = 1; k < PITCH_LAG_SPAN2; k++) {
inptr = &in[k];
const int16_t* tmp_in1 = &in[k - 1];
const int16_t* tmp_in2 = &in[PITCH_CORR_LEN2 + k - 1];
const int16_t* tmp_x = x;
int32_t tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8;
n = PITCH_CORR_LEN2;
csum32 = 0;
__asm __volatile (
".set push \n\t"
".set noreorder \n\t"
"lh %[tmp1], 0(%[tmp_in1]) \n\t"
"lh %[tmp2], 0(%[tmp_in2]) \n\t"
"mul %[tmp1], %[tmp1], %[tmp1] \n\t"
"mul %[tmp2], %[tmp2], %[tmp2] \n\t"
"srav %[tmp1], %[tmp1], %[scaling] \n\t"
"srav %[tmp2], %[tmp2], %[scaling] \n\t"
"subu %[ysum32], %[ysum32], %[tmp1] \n\t"
"bnez %[scaling], 2f \n\t"
" addu %[ysum32], %[ysum32], %[tmp2] \n\t"
"1: \n\t"
"lh %[tmp1], 0(%[inptr]) \n\t"
"lh %[tmp2], 0(%[tmp_x]) \n\t"
"lh %[tmp3], 2(%[inptr]) \n\t"
"lh %[tmp4], 2(%[tmp_x]) \n\t"
"lh %[tmp5], 4(%[inptr]) \n\t"
"lh %[tmp6], 4(%[tmp_x]) \n\t"
"lh %[tmp7], 6(%[inptr]) \n\t"
"lh %[tmp8], 6(%[tmp_x]) \n\t"
"mul %[tmp1], %[tmp1], %[tmp2] \n\t"
"mul %[tmp2], %[tmp3], %[tmp4] \n\t"
"mul %[tmp3], %[tmp5], %[tmp6] \n\t"
"mul %[tmp4], %[tmp7], %[tmp8] \n\t"
"addiu %[n], %[n], -4 \n\t"
"addiu %[inptr], %[inptr], 8 \n\t"
"addiu %[tmp_x], %[tmp_x], 8 \n\t"
"addu %[csum32], %[csum32], %[tmp1] \n\t"
"addu %[csum32], %[csum32], %[tmp2] \n\t"
"addu %[csum32], %[csum32], %[tmp3] \n\t"
"bgtz %[n], 1b \n\t"
" addu %[csum32], %[csum32], %[tmp4] \n\t"
"b 3f \n\t"
" nop \n\t"
"2: \n\t"
"lh %[tmp1], 0(%[inptr]) \n\t"
"lh %[tmp2], 0(%[tmp_x]) \n\t"
"lh %[tmp3], 2(%[inptr]) \n\t"
"lh %[tmp4], 2(%[tmp_x]) \n\t"
"lh %[tmp5], 4(%[inptr]) \n\t"
"lh %[tmp6], 4(%[tmp_x]) \n\t"
"lh %[tmp7], 6(%[inptr]) \n\t"
"lh %[tmp8], 6(%[tmp_x]) \n\t"
"mul %[tmp1], %[tmp1], %[tmp2] \n\t"
"mul %[tmp2], %[tmp3], %[tmp4] \n\t"
"mul %[tmp3], %[tmp5], %[tmp6] \n\t"
"mul %[tmp4], %[tmp7], %[tmp8] \n\t"
"addiu %[n], %[n], -4 \n\t"
"addiu %[inptr], %[inptr], 8 \n\t"
"addiu %[tmp_x], %[tmp_x], 8 \n\t"
"srav %[tmp1], %[tmp1], %[scaling] \n\t"
"srav %[tmp2], %[tmp2], %[scaling] \n\t"
"srav %[tmp3], %[tmp3], %[scaling] \n\t"
"srav %[tmp4], %[tmp4], %[scaling] \n\t"
"addu %[csum32], %[csum32], %[tmp1] \n\t"
"addu %[csum32], %[csum32], %[tmp2] \n\t"
"addu %[csum32], %[csum32], %[tmp3] \n\t"
"bgtz %[n], 2b \n\t"
" addu %[csum32], %[csum32], %[tmp4] \n\t"
"3: \n\t"
".set pop \n\t"
: [tmp1] "=&r" (tmp1), [tmp2] "=&r" (tmp2), [tmp3] "=&r" (tmp3),
[tmp4] "=&r" (tmp4), [tmp5] "=&r" (tmp5), [tmp6] "=&r" (tmp6),
[tmp7] "=&r" (tmp7), [tmp8] "=&r" (tmp8), [inptr] "+r" (inptr),
[csum32] "+r" (csum32), [tmp_x] "+r" (tmp_x), [ysum32] "+r" (ysum32),
[n] "+r" (n)
: [tmp_in1] "r" (tmp_in1), [tmp_in2] "r" (tmp_in2),
[scaling] "r" (scaling)
: "memory", "hi", "lo"
);
logcorQ8--;
lys = WebRtcIsacfix_Log2Q8((uint32_t)ysum32) >> 1; // Q8, sqrt(ysum)
if (csum32 > 0) {
lcs = WebRtcIsacfix_Log2Q8((uint32_t)csum32); // 2log(csum) in Q8
if (lcs > (lys + oneQ8)) { // csum/sqrt(ysum) > 2
*logcorQ8 = lcs - lys; // log2(csum/sqrt(ysum))
} else {
*logcorQ8 = oneQ8; // 1.00
}
} else {
*logcorQ8 = 0;
}
}
}