上周看到一个变调算法，挺有意思的，原本计划尝试用来润色TTS合成效果的。
实测感觉还需要进一步改进，待有空再思考改进方案。
算法细节原文，移步链接:
http://blogs.zynaptiq.com/bernsee/pitch-shifting-using-the-ft/
C++开源的项目，比较老的一个项目了。
源码下载地址:
http://blogs.zynaptiq.com/bernsee/download/
本人对这份算法源码进行简单的优化调整。
稍微提升了一点性能。
修改后的完整代码:
/**************************************************************************** * * NAME: smbPitchShift.cpp * VERSION: 1.2 * HOME URL: http://blogs.zynaptiq.com/bernsee * KNOWN BUGS: none * * SYNOPSIS: Routine for doing pitch shifting while maintaining * duration using the Short Time Fourier Transform. * * DESCRIPTION: The routine takes a pitchShift factor value which is between 0.5 * (one octave down) and 2. (one octave up). A value of exactly 1 does not change * the pitch. numSampsToProcess tells the routine how many samples in indata[0... * numSampsToProcess-1] should be pitch shifted and moved to outdata[0 ... * numSampsToProcess-1]. The two buffers can be identical (ie. it can process the * data in-place). fftFrameSize defines the FFT frame size used for the * processing. Typical values are 1024, 2048 and 4096. It may be any value <= * MAX_FRAME_LENGTH but it MUST be a power of 2. osamp is the STFT * oversampling factor which also determines the overlap between adjacent STFT * frames. It should at least be 4 for moderate scaling ratios. A value of 32 is * recommended for best quality. sampleRate takes the sample rate for the signal * in unit Hz, ie. 44100 for 44.1 kHz audio. The data passed to the routine in * indata[] should be in the range [-1.0, 1.0), which is also the output range * for the data, make sure you scale the data accordingly (for 16bit signed integers * you would have to divide (and multiply) by 32768). * * COPYRIGHT 1999-2015 Stephan M. Bernsee <s.bernsee [AT] zynaptiq [DOT] com> * * The Wide Open License (WOL) * * Permission to use, copy, modify, distribute and sell this software and its * documentation for any purpose is hereby granted without fee, provided that * the above copyright notice and this license appear in all source copies. * THIS SOFTWARE IS PROVIDED "AS IS" WITHOUT EXPRESS OR IMPLIED WARRANTY OF * ANY KIND. See http://www.dspguru.com/wol.htm for more information. * *****************************************************************************/ #if defined(SMBPITCHSHIFT_BUILD_DLL) #define SMTPITCHSHIFT_API __declspec(dllexport) #elif defined(SMPPITCHSHIFT_USE_DLL) #define SMTPITCHSHIFT_API __declspec(dllimport) #else #define SMTPITCHSHIFT_API #endif SMTPITCHSHIFT_API void smbPitchShift(float pitchShift, long numSampsToProcess, long fftFrameSize, long osamp, float sampleRate, float *indata, float *outdata); #define MAX_FRAME_LENGTH 8192 #include <string.h> #include <math.h> #include <stdio.h> #define M_PI 3.14159265358979323846f #define M_1_PI 0.318309886183790671538f void smbFft(float *fftBuffer, long fftFrameSize, long sign); float smbAtan2(float x, float y); // ----------------------------------------------------------------------------------------------------------------- #define FAST_MATH_TABLE_SIZE 512 const float sinTable_f32[FAST_MATH_TABLE_SIZE + 1] = { 0.00000000f, 0.01227154f, 0.02454123f, 0.03680722f, 0.04906767f, 0.06132074f, 0.07356456f, 0.08579731f, 0.09801714f, 0.11022221f, 0.12241068f, 0.13458071f, 0.14673047f, 0.15885814f, 0.17096189f, 0.18303989f, 0.19509032f, 0.20711138f, 0.21910124f, 0.23105811f, 0.24298018f, 0.25486566f, 0.26671276f, 0.27851969f, 0.29028468f, 0.30200595f, 0.31368174f, 0.32531029f, 0.33688985f, 0.34841868f, 0.35989504f, 0.37131719f, 0.38268343f, 0.39399204f, 0.40524131f, 0.41642956f, 0.42755509f, 0.43861624f, 0.44961133f, 0.46053871f, 0.47139674f, 0.48218377f, 0.49289819f, 0.50353838f, 0.51410274f, 0.52458968f, 0.53499762f, 0.54532499f, 0.55557023f, 0.56573181f, 0.57580819f, 0.58579786f, 0.59569930f, 0.60551104f, 0.61523159f, 0.62485949f, 0.63439328f, 0.64383154f, 0.65317284f, 0.66241578f, 0.67155895f, 0.68060100f, 0.68954054f, 0.69837625f, 0.70710678f, 0.71573083f, 0.72424708f, 0.73265427f, 0.74095113f, 0.74913639f, 0.75720885f, 0.76516727f, 0.77301045f, 0.78073723f, 0.78834643f, 0.79583690f, 0.80320753f, 0.81045720f, 0.81758481f, 0.82458930f, 0.83146961f, 0.83822471f, 0.84485357f, 0.85135519f, 0.85772861f, 0.86397286f, 0.87008699f, 0.87607009f, 0.88192126f, 0.88763962f, 0.89322430f, 0.89867447f, 0.90398929f, 0.90916798f, 0.91420976f, 0.91911385f, 0.92387953f, 0.92850608f, 0.93299280f, 0.93733901f, 0.94154407f, 0.94560733f, 0.94952818f, 0.95330604f, 0.95694034f, 0.96043052f, 0.96377607f, 0.96697647f, 0.97003125f, 0.97293995f, 0.97570213f, 0.97831737f, 0.98078528f, 0.98310549f, 0.98527764f, 0.98730142f, 0.98917651f, 0.99090264f, 0.99247953f, 0.99390697f, 0.99518473f, 0.99631261f, 0.99729046f, 0.99811811f, 0.99879546f, 0.99932238f, 0.99969882f, 0.99992470f, 1.00000000f, 0.99992470f, 0.99969882f, 0.99932238f, 0.99879546f, 0.99811811f, 0.99729046f, 0.99631261f, 0.99518473f, 0.99390697f, 0.99247953f, 0.99090264f, 0.98917651f, 0.98730142f, 0.98527764f, 0.98310549f, 0.98078528f, 0.97831737f, 0.97570213f, 0.97293995f, 0.97003125f, 0.96697647f, 0.96377607f, 0.96043052f, 0.95694034f, 0.95330604f, 0.94952818f, 0.94560733f, 0.94154407f, 0.93733901f, 0.93299280f, 0.92850608f, 0.92387953f, 0.91911385f, 0.91420976f, 0.90916798f, 0.90398929f, 0.89867447f, 0.89322430f, 0.88763962f, 0.88192126f, 0.87607009f, 0.87008699f, 0.86397286f, 0.85772861f, 0.85135519f, 0.84485357f, 0.83822471f, 0.83146961f, 0.82458930f, 0.81758481f, 0.81045720f, 0.80320753f, 0.79583690f, 0.78834643f, 0.78073723f, 0.77301045f, 0.76516727f, 0.75720885f, 0.74913639f, 0.74095113f, 0.73265427f, 0.72424708f, 0.71573083f, 0.70710678f, 0.69837625f, 0.68954054f, 0.68060100f, 0.67155895f, 0.66241578f, 0.65317284f, 0.64383154f, 0.63439328f, 0.62485949f, 0.61523159f, 0.60551104f, 0.59569930f, 0.58579786f, 0.57580819f, 0.56573181f, 0.55557023f, 0.54532499f, 0.53499762f, 0.52458968f, 0.51410274f, 0.50353838f, 0.49289819f, 0.48218377f, 0.47139674f, 0.46053871f, 0.44961133f, 0.43861624f, 0.42755509f, 0.41642956f, 0.40524131f, 0.39399204f, 0.38268343f, 0.37131719f, 0.35989504f, 0.34841868f, 0.33688985f, 0.32531029f, 0.31368174f, 0.30200595f, 0.29028468f, 0.27851969f, 0.26671276f, 0.25486566f, 0.24298018f, 0.23105811f, 0.21910124f, 0.20711138f, 0.19509032f, 0.18303989f, 0.17096189f, 0.15885814f, 0.14673047f, 0.13458071f, 0.12241068f, 0.11022221f, 0.09801714f, 0.08579731f, 0.07356456f, 0.06132074f, 0.04906767f, 0.03680722f, 0.02454123f, 0.01227154f, 0.00000000f, -0.01227154f, -0.02454123f, -0.03680722f, -0.04906767f, -0.06132074f, -0.07356456f, -0.08579731f, -0.09801714f, -0.11022221f, -0.12241068f, -0.13458071f, -0.14673047f, -0.15885814f, -0.17096189f, -0.18303989f, -0.19509032f, -0.20711138f, -0.21910124f, -0.23105811f, -0.24298018f, -0.25486566f, -0.26671276f, -0.27851969f, -0.29028468f, -0.30200595f, -0.31368174f, -0.32531029f, -0.33688985f, -0.34841868f, -0.35989504f, -0.37131719f, -0.38268343f, -0.39399204f, -0.40524131f, -0.41642956f, -0.42755509f, -0.43861624f, -0.44961133f, -0.46053871f, -0.47139674f, -0.48218377f, -0.49289819f, -0.50353838f, -0.51410274f, -0.52458968f, -0.53499762f, -0.54532499f, -0.55557023f, -0.56573181f, -0.57580819f, -0.58579786f, -0.59569930f, -0.60551104f, -0.61523159f, -0.62485949f, -0.63439328f, -0.64383154f, -0.65317284f, -0.66241578f, -0.67155895f, -0.68060100f, -0.68954054f, -0.69837625f, -0.70710678f, -0.71573083f, -0.72424708f, -0.73265427f, -0.74095113f, -0.74913639f, -0.75720885f, -0.76516727f, -0.77301045f, -0.78073723f, -0.78834643f, -0.79583690f, -0.80320753f, -0.81045720f, -0.81758481f, -0.82458930f, -0.83146961f, -0.83822471f, -0.84485357f, -0.85135519f, -0.85772861f, -0.86397286f, -0.87008699f, -0.87607009f, -0.88192126f, -0.88763962f, -0.89322430f, -0.89867447f, -0.90398929f, -0.90916798f, -0.91420976f, -0.91911385f, -0.92387953f, -0.92850608f, -0.93299280f, -0.93733901f, -0.94154407f, -0.94560733f, -0.94952818f, -0.95330604f, -0.95694034f, -0.96043052f, -0.96377607f, -0.96697647f, -0.97003125f, -0.97293995f, -0.97570213f, -0.97831737f, -0.98078528f, -0.98310549f, -0.98527764f, -0.98730142f, -0.98917651f, -0.99090264f, -0.99247953f, -0.99390697f, -0.99518473f, -0.99631261f, -0.99729046f, -0.99811811f, -0.99879546f, -0.99932238f, -0.99969882f, -0.99992470f, -1.00000000f, -0.99992470f, -0.99969882f, -0.99932238f, -0.99879546f, -0.99811811f, -0.99729046f, -0.99631261f, -0.99518473f, -0.99390697f, -0.99247953f, -0.99090264f, -0.98917651f, -0.98730142f, -0.98527764f, -0.98310549f, -0.98078528f, -0.97831737f, -0.97570213f, -0.97293995f, -0.97003125f, -0.96697647f, -0.96377607f, -0.96043052f, -0.95694034f, -0.95330604f, -0.94952818f, -0.94560733f, -0.94154407f, -0.93733901f, -0.93299280f, -0.92850608f, -0.92387953f, -0.91911385f, -0.91420976f, -0.90916798f, -0.90398929f, -0.89867447f, -0.89322430f, -0.88763962f, -0.88192126f, -0.87607009f, -0.87008699f, -0.86397286f, -0.85772861f, -0.85135519f, -0.84485357f, -0.83822471f, -0.83146961f, -0.82458930f, -0.81758481f, -0.81045720f, -0.80320753f, -0.79583690f, -0.78834643f, -0.78073723f, -0.77301045f, -0.76516727f, -0.75720885f, -0.74913639f, -0.74095113f, -0.73265427f, -0.72424708f, -0.71573083f, -0.70710678f, -0.69837625f, -0.68954054f, -0.68060100f, -0.67155895f, -0.66241578f, -0.65317284f, -0.64383154f, -0.63439328f, -0.62485949f, -0.61523159f, -0.60551104f, -0.59569930f, -0.58579786f, -0.57580819f, -0.56573181f, -0.55557023f, -0.54532499f, -0.53499762f, -0.52458968f, -0.51410274f, -0.50353838f, -0.49289819f, -0.48218377f, -0.47139674f, -0.46053871f, -0.44961133f, -0.43861624f, -0.42755509f, -0.41642956f, -0.40524131f, -0.39399204f, -0.38268343f, -0.37131719f, -0.35989504f, -0.34841868f, -0.33688985f, -0.32531029f, -0.31368174f, -0.30200595f, -0.29028468f, -0.27851969f, -0.26671276f, -0.25486566f, -0.24298018f, -0.23105811f, -0.21910124f, -0.20711138f, -0.19509032f, -0.18303989f, -0.17096189f, -0.15885814f, -0.14673047f, -0.13458071f, -0.12241068f, -0.11022221f, -0.09801714f, -0.08579731f, -0.07356456f, -0.06132074f, -0.04906767f, -0.03680722f, -0.02454123f, -0.01227154f, -0.00000000f }; inline float fastsin( float x) { float sinVal, fract, in; /* Temporary variables for input, output */ unsigned short index; /* Index variable */ float a, b; /* Two nearest output values */ int n; float findex; /* input x is in radians */ /* Scale the input to [0 1] range from [0 2*PI] , divide input by 2*pi */ in = x * 0.159154943092f; /* Calculation of floor value of input */ n = (int)in; /* Make negative values towards -infinity */ if (x < 0.0f) { n--; } /* Map input value to [0 1] */ in = in - (float)n; /* Calculation of index of the table */ findex = (float)FAST_MATH_TABLE_SIZE * in; if (findex >= 512.0f) { findex -= 512.0f; } index = ((unsigned short)findex) & 0x1ff; /* fractional value calculation */ fract = findex - (float)index; /* Read two nearest values of input value from the sin table */ a = sinTable_f32[index]; b = sinTable_f32[index + 1]; /* Linear interpolation process */ sinVal = (1.0f - fract)*a + fract*b; /* Return the output value */ return (sinVal); } inline float fastcos( float x) { float cosVal, fract, in; /* Temporary variables for input, output */ unsigned short index; /* Index variable */ float a, b; /* Two nearest output values */ int n; float findex; /* input x is in radians */ /* Scale the input to [0 1] range from [0 2*PI] , divide input by 2*pi, add 0.25 (pi/2) to read sine table */ in = x * 0.159154943092f + 0.25f; /* Calculation of floor value of input */ n = (int)in; /* Make negative values towards -infinity */ if (in < 0.0f) { n--; } /* Map input value to [0 1] */ in = in - (float)n; /* Calculation of index of the table */ findex = (float)FAST_MATH_TABLE_SIZE * in; index = ((unsigned short)findex) & 0x1ff; /* fractional value calculation */ fract = findex - (float)index; /* Read two nearest values of input value from the cos table */ a = sinTable_f32[index]; b = sinTable_f32[index + 1]; /* Linear interpolation process */ cosVal = (1.0f - fract)*a + fract*b; /* Return the output value */ return (cosVal); } float fastAtan2(float y, float x) { //float coeff_1 = PI/4.0f; float coeff_1 = 0.785398163f; float coeff_2 = 3.0f * coeff_1; float abs_y = abs(y) + 0.00000001f; // kludge to prevent 0/0 condition; float angle; if (x >= 0.0f) { float r = (x - abs_y) / (x + abs_y); angle = coeff_1 - coeff_1 * r; } else { float r = (x + abs_y) / (abs_y - x); angle = coeff_2 - coeff_1 * r; } return y < 0.0f ? -angle : angle; } float fastSqrt(float v) { int i; float x2, y; const float threehalfs = 1.5F; x2 = v * 0.5F; y = v; i = *(int *)&y; i = 0x5f375a86 - (i >> 1); y = *(float *)&i; y = y * (threehalfs - (x2 * y * y)); y = y * (threehalfs - (x2 * y * y)); y = y * (threehalfs - (x2 * y * y)); return v*y; } void smbPitchShift(float pitchShift, long numSampsToProcess, long fftFrameSize, long osamp, float sampleRate, float *indata, float *outdata) /* Routine smbPitchShift(). See top of file for explanation Purpose: doing pitch shifting while maintaining duration using the Short Time Fourier Transform. Author: (c)1999-2015 Stephan M. Bernsee <s.bernsee [AT] zynaptiq [DOT] com> */ { static float gInFIFO[MAX_FRAME_LENGTH]; static float gOutFIFO[MAX_FRAME_LENGTH]; static float gFFTworksp[2 * MAX_FRAME_LENGTH]; static float gLastPhase[MAX_FRAME_LENGTH / 2 + 1]; static float gSumPhase[MAX_FRAME_LENGTH / 2 + 1]; static float gOutputAccum[2 * MAX_FRAME_LENGTH]; static float gAnaFreq[MAX_FRAME_LENGTH]; static float gAnaMagn[MAX_FRAME_LENGTH]; static float gSynFreq[MAX_FRAME_LENGTH]; static float gSynMagn[MAX_FRAME_LENGTH]; static long gRover = false, gInit = false; float phase, tmp, real, imag; long i, k, qpd, index; float M_2PI = 2.0f*M_PI; float W_2PI = M_2PI / fftFrameSize; /* set up some handy variables */ long fftFrameSize2 = fftFrameSize / 2; long stepSize = fftFrameSize / osamp; float freqPerBin = sampleRate / (float)fftFrameSize; float expct = W_2PI*(float)stepSize; float pi_osamp = (M_2PI / osamp); float pi_osamp_expct = pi_osamp + expct; float pi_osamp_freqPerBin = pi_osamp / freqPerBin; float osamp_freqPerBin = osamp* freqPerBin / M_2PI; float fft_osamp = 1.0f / (fftFrameSize2*osamp); long inFifoLatency = fftFrameSize - stepSize; if (gRover == false) gRover = inFifoLatency; /* initialize our static arrays */ if (gInit == false) { memset(gInFIFO, 0, MAX_FRAME_LENGTH*sizeof(float)); memset(gOutFIFO, 0, MAX_FRAME_LENGTH*sizeof(float)); memset(gFFTworksp, 0, 2 * MAX_FRAME_LENGTH*sizeof(float)); memset(gLastPhase, 0, (MAX_FRAME_LENGTH / 2 + 1)*sizeof(float)); memset(gSumPhase, 0, (MAX_FRAME_LENGTH / 2 + 1)*sizeof(float)); memset(gOutputAccum, 0, 2 * MAX_FRAME_LENGTH*sizeof(float)); memset(gAnaFreq, 0, MAX_FRAME_LENGTH*sizeof(float)); memset(gAnaMagn, 0, MAX_FRAME_LENGTH*sizeof(float)); gInit = true; } /* main processing loop */ for (i = 0; i < numSampsToProcess; i++){ /* As long as we have not yet collected enough data just read in */ gInFIFO[gRover] = indata[i]; outdata[i] = gOutFIFO[gRover - inFifoLatency]; gRover++; /* now we have enough data for processing */ if (gRover >= fftFrameSize) { gRover = inFifoLatency; /* do windowing and re,im interleave */ float * sFFTworksp = gFFTworksp; for (k = 0; k < fftFrameSize; k++) { sFFTworksp[0] = gInFIFO[k] * (-0.5f*fastcos(W_2PI*k) + 0.5f); sFFTworksp[1] = 0.0f; sFFTworksp += 2; } /* ***************** ANALYSIS ******************* */ /* do transform */ smbFft(gFFTworksp, fftFrameSize, -1); /* this is the analysis step */ float *pFFTworksp = gFFTworksp; for (k = 0; k <= fftFrameSize2; k++) { /* de-interlace FFT buffer */ real = pFFTworksp[0]; imag = pFFTworksp[1]; /* compute magnitude and phase */ gAnaMagn[k] = 2.0f*fastSqrt(real*real + imag*imag); phase = fastAtan2(imag, real); /* compute phase difference */ /* subtract expected phase difference */ tmp = (phase - gLastPhase[k]) - (float)k*expct; gLastPhase[k] = phase; /* map delta phase into +/- Pi interval */ qpd = tmp * M_1_PI; if (qpd >= 0) qpd += qpd & 1; else qpd -= qpd & 1; /* get deviation from bin frequency from the +/- Pi interval */ /* compute the k-th partials' true frequency */ /* store magnitude and true frequency in analysis arrays */ gAnaFreq[k] = (k + (tmp - M_PI*qpd) * osamp_freqPerBin); pFFTworksp += 2; } /* ***************** PROCESSING ******************* */ /* this does the actual pitch shifting */ memset(gSynMagn, 0, fftFrameSize*sizeof(float)); memset(gSynFreq, 0, fftFrameSize*sizeof(float)); for (k = 0; k <= fftFrameSize2; k++) { index = k*pitchShift; if (index <= fftFrameSize2) { gSynMagn[index] += gAnaMagn[k]; gSynFreq[index] = gAnaFreq[k] * pitchShift; } } /* ***************** SYNTHESIS ******************* */ /* this is the synthesis step */ float *tFFTworksp = gFFTworksp; for (k = 0; k <= fftFrameSize2; k++) { gSumPhase[k] += pi_osamp_freqPerBin * gSynFreq[k] - pi_osamp_expct*k; tFFTworksp[0] = gSynMagn[k] * fastcos(gSumPhase[k]); tFFTworksp[1] = gSynMagn[k] * fastsin(gSumPhase[k]); tFFTworksp += 2; } /* zero negative frequencies */ memset(gFFTworksp + (fftFrameSize + 2), 0, sizeof(float) * 2 * fftFrameSize); /* do inverse transform */ smbFft(gFFTworksp, fftFrameSize, 1); /* do windowing and add to output accumulator */ float *ppFFTworksp = gFFTworksp; for (k = 0; k < fftFrameSize; k++) { gOutputAccum[k] += (-fastcos(W_2PI*k) *ppFFTworksp[0] * fft_osamp) + ppFFTworksp[0] * fft_osamp; ppFFTworksp += 2; } memcpy(gOutFIFO, gOutputAccum, stepSize*sizeof(float)); /* shift accumulator */ memmove(gOutputAccum, gOutputAccum + stepSize, fftFrameSize*sizeof(float)); /* move input FIFO */ memcpy(gInFIFO, gInFIFO + stepSize, sizeof(float)*inFifoLatency); } } } // ----------------------------------------------------------------------------------------------------------------- void smbFft(float *fftBuffer, long fftFrameSize, long sign) /* FFT routine, (C)1996 S.M.Bernsee. Sign = -1 is FFT, 1 is iFFT (inverse) Fills fftBuffer[0...2*fftFrameSize-1] with the Fourier transform of the time domain data in fftBuffer[0...2*fftFrameSize-1]. The FFT array takes and returns the cosine and sine parts in an interleaved manner, ie. fftBuffer[0] = cosPart[0], fftBuffer[1] = sinPart[0], asf. fftFrameSize must be a power of 2. It expects a complex input signal (see footnote 2), ie. when working with 'common' audio signals our input signal has to be passed as {in[0],0.,in[1],0.,in[2],0.,...} asf. In that case, the transform of the frequencies of interest is in fftBuffer[0...fftFrameSize]. */ { float wr, wi, arg, *p1, *p2, temp; float tr, ti, ur, ui, *p1r, *p1i, *p2r, *p2i; long i, bitm, j, le, le2, k; for (i = 2; i < 2 * fftFrameSize - 2; i += 2) { for (bitm = 2, j = 0; bitm < 2 * fftFrameSize; bitm <<= 1) { if (i & bitm) j++; j <<= 1; } if (i < j) { p1 = fftBuffer + i; p2 = fftBuffer + j; temp = *p1; *(p1++) = *p2; *(p2++) = temp; temp = *p1; *p1 = *p2; *p2 = temp; } } long k_size = (long)(log((float)fftFrameSize) / log(2.0f) + 0.5f); for (k = 0, le = 2; k < k_size; k++) { le <<= 1; le2 = le >> 1; ur = 1.0f; ui = 0.0f; arg = M_PI / (le2 >> 1); wr = fastcos(arg); wi = sign*fastsin(arg); for (j = 0; j < le2; j += 2) { p1r = fftBuffer + j; p1i = p1r + 1; p2r = p1r + le2; p2i = p2r + 1; for (i = j; i < 2 * fftFrameSize; i += le) { tr = *p2r * ur - *p2i * ui; ti = *p2r * ui + *p2i * ur; *p2r = *p1r - tr; *p2i = *p1i - ti; *p1r += tr; *p1i += ti; p1r += le; p1i += le; p2r += le; p2i += le; } tr = ur*wr - ui*wi; ui = ur*wi + ui*wr; ur = tr; } } } // ----------------------------------------------------------------------------------------------------------------- /* 12/12/02, smb PLEASE NOTE: There have been some reports on domain errors when the atan2() function was used as in the above code. Usually, a domain error should not interrupt the program flow (maybe except in Debug mode) but rather be handled "silently" and a global variable should be set according to this error. However, on some occasions people ran into this kind of scenario, so a replacement atan2() function is provided here. If you are experiencing domain errors and your program stops, simply replace all instances of atan2() with calls to the smbAtan2() function below. */ float smbAtan2(float x, float y) { float signx; if (x > 0.0f) signx = 1.0f; else signx = -1.0f; if (x == 0.0f) return 0.0f; if (y == 0.0f) return signx * M_PI / 2.0f; return fastAtan2(x, y); }
主要优化处理是:
sin改为fastsin
cos改为fastcos
atan2 改为fastAtan2
sqrt改为fastSqrt
还有其他一些逻辑上的调整。
附上完整示例代码:
#define _CRT_SECURE_NO_WARNINGS #define _CRT_SECURE_NO_DEPRECATE 1 #define _CRT_NONSTDC_NO_DEPRECATE 1 #include <stdio.h> #include <stdlib.h> #include <stdint.h> #include <time.h> #include <iostream> //采用https://github.com/mackron/dr_libs/blob/master/dr_wav.h 解码 #define DR_WAV_IMPLEMENTATION #include "dr_wav.h" //采用http://blogs.zynaptiq.com/bernsee/pitch-shifting-using-the-ft/ #include "smbPitchShift.h" auto const epoch = clock(); static double now() { return (clock() - epoch); }; template <typename FN> static double bench(const FN &fn) { auto took = -now(); return (fn(), took + now()) / 1000; } //写wav文件 void wavWrite_float(char* filename, float* buffer, int sampleRate, uint32_t totalSampleCount) { FILE* fp = fopen(filename, "wb"); if (fp == NULL) { printf("文件打开失败. "); return; } //修正写入的buffer长度 totalSampleCount *= sizeof(float); int nbit = 32; int FORMAT_PCM = DR_WAVE_FORMAT_IEEE_FLOAT; int nbyte = nbit / 8; char text[4] = { 'R', 'I', 'F', 'F' }; uint32_t long_number = 36 + totalSampleCount; fwrite(text, 1, 4, fp); fwrite(&long_number, 4, 1, fp); text[0] = 'W'; text[1] = 'A'; text[2] = 'V'; text[3] = 'E'; fwrite(text, 1, 4, fp); text[0] = 'f'; text[1] = 'm'; text[2] = 't'; text[3] = ' '; fwrite(text, 1, 4, fp); long_number = 16; fwrite(&long_number, 4, 1, fp); int16_t short_number = FORMAT_PCM;//默认音频格式 fwrite(&short_number, 2, 1, fp); short_number = 1; // 音频通道数 fwrite(&short_number, 2, 1, fp); long_number = sampleRate; // 采样率 fwrite(&long_number, 4, 1, fp); long_number = sampleRate * nbyte; // 比特率 fwrite(&long_number, 4, 1, fp); short_number = nbyte; // 块对齐 fwrite(&short_number, 2, 1, fp); short_number = nbit; // 采样精度 fwrite(&short_number, 2, 1, fp); char data[4] = { 'd', 'a', 't', 'a' }; fwrite(data, 1, 4, fp); long_number = totalSampleCount; fwrite(&long_number, 4, 1, fp); fwrite(buffer, totalSampleCount, 1, fp); fclose(fp); } //读取wav文件 float* wavRead_float(char* filename, uint32_t* sampleRate, uint64_t *totalSampleCount) { unsigned int channels; float* buffer = drwav_open_and_read_file_f32(filename, &channels, sampleRate, totalSampleCount); if (buffer == NULL) { printf("读取wav文件失败."); } //仅仅处理单通道音频 if (channels != 1) { drwav_free(buffer); buffer = NULL; *sampleRate = 0; *totalSampleCount = 0; } return buffer; } //分割路径函数 void splitpath(const char* path, char* drv, char* dir, char* name, char* ext) { const char* end; const char* p; const char* s; if (path[0] && path[1] == ':') { if (drv) { *drv++ = *path++; *drv++ = *path++; *drv = ' '; } } else if (drv) *drv = ' '; for (end = path; *end && *end != ':';) end++; for (p = end; p > path && *--p != '\' && *p != '/';) if (*p == '.') { end = p; break; } if (ext) for (s = end; (*ext = *s++);) ext++; for (p = end; p > path;) if (*--p == '\' || *p == '/') { p++; break; } if (name) { for (s = p; s < end;) *name++ = *s++; *name = ' '; } if (dir) { for (s = path; s < p;) *dir++ = *s++; *dir = ' '; } } int main(int argc, char* argv[]) { std::cout << "Audio Processing " << std::endl; std::cout << "博客:http://cpuimage.cnblogs.com/" << std::endl; std::cout << "支持解析单通道wav格式的变调处理." << std::endl; if (argc < 2) return -1; char* in_file = argv[1]; //音频采样率 uint32_t sampleRate = 0; //总音频采样数 uint64_t totalSampleCount = 0; float* wavBuffer = NULL; double nLoadTime = bench([&] { wavBuffer = wavRead_float(in_file, &sampleRate, &totalSampleCount); }); std::cout << " 加载耗时: " << int(nLoadTime * 1000) << " 毫秒" << std::endl; //如果加载成功 if (wavBuffer != NULL) { float semitones = 8; // 向上移动8个半音 float pitchShift = pow(2.0f, semitones / 12.0f); //将半音转换为因子 printf("pitchShift:%f", pitchShift); double nProcessTime = bench([&] { smbPitchShift(pitchShift, totalSampleCount, 2048, 4, sampleRate, wavBuffer, wavBuffer); }); std::cout << " 处理耗时: " << int(nProcessTime * 1000) << " 毫秒" << std::endl; } //保存结果 double nSaveTime = bench([&] { char drive[3]; char dir[256]; char fname[256]; char ext[256]; char out_file[1024]; splitpath(in_file, drive, dir, fname, ext); sprintf(out_file, "%s%s%s_out%s", drive, dir, fname, ext); wavWrite_float(out_file, wavBuffer, sampleRate, totalSampleCount); }); std::cout << " 保存耗时: " << int(nSaveTime * 1000) << " 毫秒" << std::endl; if (wavBuffer) { free(wavBuffer); } getchar(); std::cout << "按任意键退出程序 " << std::endl; return 0; }
示例具体流程为：
加载wav(拖放wav文件到可执行文件上)->变调处理->保存wav
并对 加载，处理，保存 这3个环节都进行了耗时计算并输出。
其中主要的参数是 float semitones = 8，数值越高音调越高。
参数为8时，变调出来的声音有点像汤姆猫。
注：示例代码仅支持处理单通道音频，多通道稍做改动即可支持。
若有其他相关问题或者需求也可以邮件联系俺探讨。
邮箱地址是:
gaozhihan@vip.qq.com
若此博文能帮到您,欢迎扫码小额赞助。
微信:
支付宝:
