Menu
▾
▴
[OMPT-svn-commit] SF.net SVN: modplug:[3705] trunk/OpenMPT
|
From: <sag...@us...> - 2014-02-12 15:24:48
|
Revision: 3705
http://sourceforge.net/p/modplug/code/3705
Author: saga-games
Date: 2014-02-12 15:24:37 +0000 (Wed, 12 Feb 2014)
Log Message:
-----------
[Mod] Update smbPitchShift to v1.2
[Mod] Use better quality default for pitch shifting.
[Mod] Fix 8-bit conversion code for pitch shifting so that upsampling is no longer required.
[Mod] OpenMPT: Version is now 1.22.07.21
Modified Paths:
--------------
trunk/OpenMPT/common/versionNumber.h
trunk/OpenMPT/include/smbPitchShift/smbPitchShift.cpp
trunk/OpenMPT/include/smbPitchShift/smbPitchShift.h
trunk/OpenMPT/mptrack/Ctrl_smp.cpp
trunk/OpenMPT/soundlib/SampleFormatConverters.h
Modified: trunk/OpenMPT/common/versionNumber.h
===================================================================
--- trunk/OpenMPT/common/versionNumber.h 2014-02-12 13:59:49 UTC (rev 3704)
+++ trunk/OpenMPT/common/versionNumber.h 2014-02-12 15:24:37 UTC (rev 3705)
@@ -17,7 +17,7 @@
#define VER_MAJORMAJOR 1
#define VER_MAJOR 22
#define VER_MINOR 07
-#define VER_MINORMINOR 20
+#define VER_MINORMINOR 21
//Version string. For example "1.17.02.28"
#define MPT_VERSION_STR VER_STRINGIZE(VER_MAJORMAJOR) "." VER_STRINGIZE(VER_MAJOR) "." VER_STRINGIZE(VER_MINOR) "." VER_STRINGIZE(VER_MINORMINOR)
Modified: trunk/OpenMPT/include/smbPitchShift/smbPitchShift.cpp
===================================================================
--- trunk/OpenMPT/include/smbPitchShift/smbPitchShift.cpp 2014-02-12 13:59:49 UTC (rev 3704)
+++ trunk/OpenMPT/include/smbPitchShift/smbPitchShift.cpp 2014-02-12 15:24:37 UTC (rev 3705)
@@ -1,7 +1,7 @@
/****************************************************************************
*
* NAME: smbPitchShift.cpp
-* VERSION: 1.1
+* VERSION: 1.2
* HOME URL: http://www.dspdimension.com
* KNOWN BUGS: none
*
@@ -15,7 +15,7 @@
* 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_FFT_FRAME_LENGTH but it MUST be a power of 2. osamp is the STFT
+* 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
@@ -24,7 +24,7 @@
* for the data, make sure you scale the data accordingly (for 16bit signed integers
* you would have to divide (and multiply) by 32768).
*
-* COPYRIGHT 1999-2003 Stephan M. Bernsee <sm...@ds...>
+* COPYRIGHT 1999-2009 Stephan M. Bernsee <smb [AT] dspdimension [DOT] com>
*
* The Wide Open License (WOL)
*
@@ -40,8 +40,10 @@
#include <string.h>
#include <math.h>
-#include "smbPitchShift.h"
+#include <stdio.h>
+#define M_PI 3.14159265358979323846
+#define MAX_FRAME_LENGTH 8192
void smbFft(float *fftBuffer, long fftFrameSize, long sign);
double smbAtan2(double x, double y);
@@ -50,14 +52,15 @@
// -----------------------------------------------------------------------------------------------------------------
-void smbPitchShift(float pitchShift, long numSampsToProcess, long fftFrameSize, long osamp, float sampleRate, float *indata, float *outdata, bool gInit)
+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)1999-2002 Stephan M. Bernsee <sm...@ds...>
+ Author: (c)1999-2009 Stephan M. Bernsee <smb [AT] dspdimension [DOT] com>
*/
{
+
static float gInFIFO[MAX_FRAME_LENGTH];
static float gOutFIFO[MAX_FRAME_LENGTH];
static float gFFTworksp[2*MAX_FRAME_LENGTH];
@@ -68,34 +71,32 @@
static float gAnaMagn[MAX_FRAME_LENGTH];
static float gSynFreq[MAX_FRAME_LENGTH];
static float gSynMagn[MAX_FRAME_LENGTH];
- static long gRover = false;
+ static long gRover = false, gInit = false;
double magn, phase, tmp, window, real, imag;
double freqPerBin, expct;
- long i,k, qpd, index, inFifoLatency, stepSize, fftFrameSize2, fadeZoneLen;
+ long i,k, qpd, index, inFifoLatency, stepSize, fftFrameSize2;
+ /* set up some handy variables */
+ fftFrameSize2 = fftFrameSize/2;
+ stepSize = fftFrameSize/osamp;
+ freqPerBin = sampleRate/(double)fftFrameSize;
+ expct = 2.*M_PI*(double)stepSize/(double)fftFrameSize;
+ inFifoLatency = fftFrameSize-stepSize;
+ if (gRover == false) gRover = inFifoLatency;
+
/* initialize our static arrays */
- if (gInit == true) {
+ 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*sizeof(float)/2);
- memset(gSumPhase, 0, MAX_FRAME_LENGTH*sizeof(float)/2);
+ 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));
- gRover = false;
- gInit = false;
+ gInit = true;
}
- /* set up some handy variables */
- fadeZoneLen = fftFrameSize/2;
- fftFrameSize2 = fftFrameSize/2;
- stepSize = fftFrameSize/osamp;
- freqPerBin = sampleRate/(double)fftFrameSize;
- expct = 2.*M_PI*(double)stepSize/(double)fftFrameSize;
- inFifoLatency = fftFrameSize-stepSize;
- if (gRover == false) gRover = inFifoLatency;
-
/* main processing loop */
for (i = 0; i < numSampsToProcess; i++){
@@ -156,20 +157,18 @@
}
-
-
/* ***************** 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[k] += gAnaMagn[index];
- gSynFreq[k] = gAnaFreq[index] * pitchShift;
- }
+ 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 */
for (k = 0; k <= fftFrameSize2; k++) {
@@ -253,7 +252,7 @@
*p1 = *p2; *p2 = temp;
}
}
- for (k = 0, le = 2; k < (long)(log((double)fftFrameSize)/log(2.)); k++) {
+ for (k = 0, le = 2; k < (long)(log((double)fftFrameSize)/log(2.)+.5); k++) {
le <<= 1;
le2 = le>>1;
ur = 1.0;
Modified: trunk/OpenMPT/include/smbPitchShift/smbPitchShift.h
===================================================================
--- trunk/OpenMPT/include/smbPitchShift/smbPitchShift.h 2014-02-12 13:59:49 UTC (rev 3704)
+++ trunk/OpenMPT/include/smbPitchShift/smbPitchShift.h 2014-02-12 15:24:37 UTC (rev 3705)
@@ -1,9 +1,7 @@
#ifndef SMBPITCHSHIFT_H
#define SMBPITCHSHIFT_H
-#define M_PI 3.14159265358979323846
+extern void smbPitchShift(float pitchShift, long numSampsToProcess, long fftFrameSize, long osamp, float sampleRate, float *indata, float *outdata);
#define MAX_FRAME_LENGTH 8192
-extern void smbPitchShift(float pitchShift, long numSampsToProcess, long fftFrameSize, long osamp, float sampleRate, float *indata, float *outdata, bool gInit);
-
#endif
\ No newline at end of file
Modified: trunk/OpenMPT/mptrack/Ctrl_smp.cpp
===================================================================
--- trunk/OpenMPT/mptrack/Ctrl_smp.cpp 2014-02-12 13:59:49 UTC (rev 3704)
+++ trunk/OpenMPT/mptrack/Ctrl_smp.cpp 2014-02-12 15:24:37 UTC (rev 3705)
@@ -30,6 +30,7 @@
#include "../common/StringFixer.h"
#include "MemoryMappedFile.h"
#include "../soundlib/FileReader.h"
+#include "../soundlib/SampleFormatConverters.h"
#include <Shlwapi.h>
#include "FileDialog.h"
@@ -46,26 +47,27 @@
// Float point comparison
bool EqualTof(const float a, const float b)
{
- return (fabs(a - b) <= FLOAT_ERROR);
+ return (fabs(a - b) <= FLOAT_ERROR);
}
// Round floating point value to "digit" number of digits
float Round(const float value, const int digit)
{
float v = 0.1f * (value * powf(10.0f, (float)(digit + 1)) + (value < 0.0f ? -5.0f : 5.0f));
- modff(v, &v);
- return v / powf(10.0f, (float)digit);
+ modff(v, &v);
+ return v / powf(10.0f, (float)digit);
}
-// Deduce exponent from equation : v = 2^exponent
-int PowerOf2Exponent(const unsigned int v)
+template<int v>
+int PowerOf2Exponent()
{
- return (int)_logb((double)v);
- //float v2f = (float)v;
- //return ( (*(int *)&v2f >> 23) & 0xff ) - 127;
+ if(v <= 1)
+ return 0;
+ else
+ return 1 + PowerOf2Exponent<v / 2>();
}
-// -! TEST#0029
+
#define BASENOTE_MIN (1*12) // C-1
#define BASENOTE_MAX (9*12) // C-9
@@ -284,15 +286,16 @@
// Quality selection
combo = (CComboBox *)GetDlgItem(IDC_COMBO5);
- if(combo){
+ if(combo)
+ {
// Allow quality from 4 to 128
for(int i = 4 ; i <= 128 ; i++)
{
wsprintf(str,"%d",i);
combo->SetItemData(combo->AddString(str), i-4);
}
- // Set 4 as default quality
- combo->SetCurSel(0);
+ // Set 32 as default quality
+ combo->SetCurSel(32 - 4);
}
// FFT size selection
@@ -300,14 +303,15 @@
if(combo)
{
// Deduce exponent from equation : MAX_FRAME_LENGTH = 2^exponent
- int exponent = PowerOf2Exponent(MAX_FRAME_LENGTH);
+ const int exponent = PowerOf2Exponent<MAX_FRAME_LENGTH>();
// Allow FFT size from 2^8 (256) to 2^exponent (MAX_FRAME_LENGTH)
- for(int i = 8 ; i <= exponent ; i++){
+ for(int i = 8 ; i <= exponent ; i++)
+ {
wsprintf(str,"%d",1<<i);
combo->SetItemData(combo->AddString(str), i-8);
}
- // Set 2048 as default FFT size
- combo->SetCurSel(3);
+ // Set 4096 as default FFT size
+ combo->SetCurSel(4);
}
// Stretch ratio
@@ -1163,9 +1167,9 @@
m_modDoc.GetSampleUndo().PrepareUndo(iSmp, sundo_update, "Normalize", selStart, selEnd);
- if(sample.uFlags & CHN_STEREO) { selStart *= 2; selEnd *= 2; }
+ if(sample.uFlags[CHN_STEREO]) { selStart *= 2; selEnd *= 2; }
- if(sample.uFlags & CHN_16BIT)
+ if(sample.uFlags[CHN_16BIT])
{
int16 *p = (int16 *)sample.pSample;
int max = 1;
@@ -1234,10 +1238,10 @@
m_modDoc.GetSampleUndo().PrepareUndo(m_nSample, sundo_update, "Amplify", selection.nStart, selection.nEnd);
- if (sample.uFlags & CHN_STEREO) { selection.nStart *= 2; selection.nEnd *= 2; }
+ if (sample.uFlags[CHN_STEREO]) { selection.nStart *= 2; selection.nEnd *= 2; }
SmpLength len = selection.nEnd - selection.nStart;
if ((bFadeIn) && (bFadeOut)) lAmp *= 4;
- if (sample.uFlags & CHN_16BIT)
+ if (sample.uFlags[CHN_16BIT])
{
signed short *p = ((signed short *)sample.pSample) + selection.nStart;
@@ -1426,7 +1430,7 @@
{
SmpLength len = dwEnd - dwStart;
int maxndx = sample.nLength;
- if (sample.uFlags & CHN_16BIT)
+ if (sample.uFlags[CHN_16BIT])
{
signed short *psrc = ((signed short *)pOriginal)+iCh;
signed short *pdest = ((signed short *)pNewSample)+dwStart*nCh+iCh;
@@ -1563,7 +1567,7 @@
{
int len = dwRemove;
int maxndx = sample.nLength;
- if (sample.uFlags & CHN_16BIT)
+ if (sample.uFlags[CHN_16BIT])
{
signed short *psrc = ((signed short *)pOriginal)+iCh;
signed short *pdest = ((signed short *)pNewSample)+dwStart*nCh+iCh;
@@ -1701,7 +1705,7 @@
CPSRatioCalc dlg(m_sndFile, m_nSample, m_dTimeStretchRatio, this);
if (dlg.DoModal() != IDOK) return;
- //Update ratio value&textbox
+ //Update ratio value&textbox
m_dTimeStretchRatio = dlg.m_dRatio;
UpdateData(FALSE);
//end rewbs.timeStretchMods
@@ -1994,6 +1998,7 @@
return 0;
}
+
int CCtrlSamples::PitchShift(float pitch)
//---------------------------------------
{
@@ -2005,8 +2010,8 @@
ModSample &sample = m_sndFile.GetSample(m_nSample);
// Get number of channels & sample size
- BYTE smpsize = sample.GetElementarySampleSize();
- UINT nChn = sample.GetNumChannels();
+ const uint8 smpsize = sample.GetElementarySampleSize();
+ const uint8 nChn = sample.GetNumChannels();
// Get selected oversampling - quality - (also refered as FFT overlapping) factor
CComboBox *combo = (CComboBox *)GetDlgItem(IDC_COMBO5);
@@ -2038,33 +2043,21 @@
// Show wait mouse cursor
BeginWaitCursor();
-
- // PitchShift seems to work only with 16-bit samples.
- if(smpsize != 2)
- {
- // This has to be converted to 16-bit first.
- SetSelectionPoints(0, 0); // avoid partial upsampling.
- OnUpsample();
- smpsize = sample.GetElementarySampleSize();
- }
-
// Get original sample rate
- long lSampleRate = sample.GetSampleRate(m_sndFile.GetType());
+ const float sampleRate = static_cast<float>(sample.GetSampleRate(m_sndFile.GetType()));
- // Deduce max sample value (float conversion step)
- float maxSampleValue = float(( 1 << (smpsize * 8 - 1) ) - 1);
+ // Allocate working buffer
+ const size_t bufferSize = MAX_BUFFER_LENGTH + fft;
+ float *buffer = new float[bufferSize];
- // Allocate working buffers
- float * buffer = new float[MAX_BUFFER_LENGTH + fft];
- float * outbuf = new float[MAX_BUFFER_LENGTH + fft];
-
m_modDoc.GetSampleUndo().PrepareUndo(m_nSample, sundo_replace, "Pitch Shift");
// Process each channel separately
- for(UINT i = 0 ; i < nChn ; i++){
+ for(uint8 i = 0 ; i < nChn ; i++)
+ {
- UINT pos = 0;
- UINT len = MAX_BUFFER_LENGTH;
+ SmpLength pos = 0;
+ SmpLength len = MAX_BUFFER_LENGTH;
// Process sample buffer using MAX_BUFFER_LENGTH (max) sized chunk steps (in order to allow
// the processing of BIG samples...)
@@ -2077,16 +2070,16 @@
// TRICK : output buffer offset management
// as the pitch-shifter adds some blank signal in head of output buffer (matching FFT
// length - in short it needs a certain amount of data before being able to output some
- // meaningfull processed samples) , in order to avoid this behaviour , we will ignore
+ // meaningfull processed samples) , in order to avoid this behaviour , we will ignore
// the first FFT_length samples and process the same amount of extra blank samples
// (all 0.0f) at the end of the buffer (those extra samples will benefit from internal
// FFT data computed during the previous steps resulting in a correct and consistent
// signal output).
bool bufstart = ( pos == 0 );
bool bufend = ( pos + MAX_BUFFER_LENGTH >= sample.nLength );
- UINT startoffset = ( bufstart ? fft : 0 );
- UINT inneroffset = ( bufstart ? 0 : fft );
- UINT finaloffset = ( bufend ? fft : 0 );
+ SmpLength startoffset = ( bufstart ? fft : 0 );
+ SmpLength inneroffset = ( bufstart ? 0 : fft );
+ SmpLength finaloffset = ( bufend ? fft : 0 );
// Show progress bar using process button painting & text label
CHAR progress[16];
@@ -2103,49 +2096,43 @@
::GdiFlush();
// Re-initialize pitch-shifter with blank FFT before processing 1st chunk of current channel
- if(bufstart){
- for(UINT j = 0 ; j < fft ; j++) buffer[j] = 0.0f;
- smbPitchShift(pitch, fft, fft, ovs, (float)lSampleRate, buffer, outbuf, true);
+ if(bufstart)
+ {
+ for(UINT j = 0; j < fft; j++) buffer[j] = 0.0f;
+ smbPitchShift(pitch, fft, fft, ovs, sampleRate, buffer, buffer);
}
// Convert current channel's data chunk to float
- BYTE * ptr = (BYTE *)sample.pSample + pos * smpsize * nChn + i * smpsize;
+ int8 *ptr = (int8 *)sample.pSample + pos * smpsize * nChn + i * smpsize;
- for(UINT j = 0 ; j < len ; j++){
- switch(smpsize){
- case 2:
- buffer[j] = ((float)(*(SHORT *)ptr)) / maxSampleValue;
- break;
- case 1:
- buffer[j] = ((float)*ptr) / maxSampleValue;
- break;
- }
- ptr += smpsize * nChn;
+ switch(smpsize)
+ {
+ case 1:
+ CopySample<SC::ConversionChain<SC::Convert<float, int8>, SC::DecodeIdentity<int8> > >(buffer, len, 1, ptr, sizeof(int8) * len, nChn);
+ break;
+ case 2:
+ CopySample<SC::ConversionChain<SC::Convert<float, int16>, SC::DecodeIdentity<int16> > >(buffer, len, 1, (int16 *)ptr, sizeof(int16) * len, nChn);
+ break;
}
// Fills extra blank samples (read TRICK description comment above)
- if(bufend) for(UINT j = len ; j < len + finaloffset ; j++) buffer[j] = 0.0f;
+ if(bufend) for(SmpLength j = len ; j < len + finaloffset ; j++) buffer[j] = 0.0f;
// Apply pitch shifting
- smbPitchShift(pitch, len + finaloffset, fft, ovs, (float)lSampleRate, buffer, outbuf, false);
+ smbPitchShift(pitch, static_cast<long>(len + finaloffset), fft, ovs, sampleRate, buffer, buffer);
// Restore pitched-shifted float sample into original sample buffer
- ptr = (BYTE *)sample.pSample + (pos - inneroffset) * smpsize * nChn + i * smpsize;
+ ptr = (int8 *)sample.pSample + (pos - inneroffset) * smpsize * nChn + i * smpsize;
+ const SmpLength copyLength = len + finaloffset - startoffset + 1;
- for(UINT j = startoffset ; j < len + finaloffset ; j++){
- // Just perform a little bit of clipping...
- float v = outbuf[j];
- CLIP_SOUND(v);
- // ...before converting back to buffer
- switch(smpsize){
- case 2:
- *(SHORT *)ptr = (SHORT)(v * maxSampleValue);
- break;
- case 1:
- *ptr = (BYTE)(v * maxSampleValue);
- break;
- }
- ptr += smpsize * nChn;
+ switch(smpsize)
+ {
+ case 1:
+ CopySample<SC::ConversionChain<SC::Convert<int8, float>, SC::DecodeIdentity<float> > >((int8 *)ptr, copyLength, nChn, buffer + startoffset, sizeof(float) * bufferSize, 1);
+ break;
+ case 2:
+ CopySample<SC::ConversionChain<SC::Convert<int16, float>, SC::DecodeIdentity<float> > >((int16 *)ptr, copyLength, nChn, buffer + startoffset, sizeof(float) * bufferSize, 1);
+ break;
}
// Next buffer chunk
@@ -2153,9 +2140,8 @@
}
}
- // Free working buffers
+ // Free working buffer
delete [] buffer;
- delete [] outbuf;
// Free progress bar brushes
DeleteObject((HBRUSH)green);
@@ -2412,10 +2398,12 @@
int nPan = GetDlgItemInt(IDC_EDIT9);
if (nPan < 0) nPan = 0;
//rewbs.fix36944: sample pan range to 0-64.
- if (m_sndFile.m_nType == MOD_TYPE_XM) {
- if (nPan>255) nPan = 255; // displayed panning will be 0-255 with XM
- } else {
- if (nPan>64) nPan = 64; // displayed panning will be 0-64 with anything but XM.
+ if (m_sndFile.GetType() == MOD_TYPE_XM)
+ {
+ if (nPan > 255) nPan = 255; // displayed panning will be 0-255 with XM
+ } else
+ {
+ if (nPan > 64) nPan = 64; // displayed panning will be 0-64 with anything but XM.
nPan = nPan << 2; // so we x4 to get MPT's internal 0-256 range.
}
//end rewbs.fix36944
Modified: trunk/OpenMPT/soundlib/SampleFormatConverters.h
===================================================================
--- trunk/OpenMPT/soundlib/SampleFormatConverters.h 2014-02-12 13:59:49 UTC (rev 3704)
+++ trunk/OpenMPT/soundlib/SampleFormatConverters.h 2014-02-12 15:24:37 UTC (rev 3705)
@@ -362,6 +362,28 @@
};
template <>
+struct Convert<float32, int8>
+{
+ typedef int8 input_t;
+ typedef float32 output_t;
+ forceinline output_t operator() (input_t val)
+ {
+ return val * (1.0f / static_cast<float>((unsigned int)1<<7));
+ }
+};
+
+template <>
+struct Convert<float32, int16>
+{
+ typedef int16 input_t;
+ typedef float32 output_t;
+ forceinline output_t operator() (input_t val)
+ {
+ return val * (1.0f / static_cast<float>((unsigned int)1<<15));
+ }
+};
+
+template <>
struct Convert<float32, int32>
{
typedef int32 input_t;
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
