diff --git a/include/Release.htm b/include/Release.htm
index fe58c8d..2064c6c 100644
--- a/include/Release.htm
+++ b/include/Release.htm
@@ -1,4 +1,4 @@
-
+
@@ -25,9 +25,17 @@
exhale - ecodis extended high-efficiency and low-complexity encoder
Software Release Notes, Version History, Known Issues, Upcoming Feature Roadmap
- The version of this distribution of the «exhale» software release is 1.2.1 (official public minor release) from December 2023. Please check www.ecodis.de regularly for new versions of this software. A summary of each version up to this release, a list of known issues with this release, and a roadmap of additional functionality are provided below.
+ The version of this distribution of the «exhale» software release is 1.2.1.2 (official public bugfix release) from May 2025. Please check www.ecodis.de regularly for new versions of this software. A summary of each version up to this release, a list of known issues with this release, and a roadmap of additional functionality are provided below.
Chronological Version History
- Version 1.2.1 Dec. 2023, this release
+ Version 1.2.1.2 May 2025, this release
+
+ exhaleApp: avoid out of bounds AU writeouts, fix 5.1 WAV/MPEG channel mappingexhaleLib: relax max. quantized spectral magnitude restriction1, fix MSE test codeexhaleLib: clean quantizer+entropy coder, increase max. IPF AU size to 12 kbit/chexhaleLib: fix more high-rate efficiency issues, fix mode-f stereo coding (issue 33)
+1: optional, activated via define SFB_QUANT_PERCEPT_OPT = 1 or 0, triggers FFmpeg decoder bug!
+ Version 1.2.1 Dec. 2023
exhaleApp: always determine MPEG-4 instantaneous bit-rate across 2048 samplesexhaleLib: added code for MSE optimized encoding (for tests, disabled by default)
Roadmap of Upcoming Features
If you are in need of an additional library or application feature not mentioned below, please contact ecodis or a contributor with a request, and we will see what we can do.
- exhaleLib: finalization of support for 5.1–7.1 multichannel coding, no version planexhaleLib: speed-ups and further quality tuning for difficult signals, as necessary.
-Written by C. R. Helmrich for exhale 1.2.1, Dec. 2023. Available at www.ecodis.de/exhale/release.htm.
+Written by C. R. Helmrich for exhale 1.2.2, Dec. 2025. Available at www.ecodis.de/exhale/release.htm.
|
diff --git a/src/lib/exhaleEnc.cpp b/src/lib/exhaleEnc.cpp
index 4293292..e7438d1 100644
--- a/src/lib/exhaleEnc.cpp
+++ b/src/lib/exhaleEnc.cpp
@@ -294,7 +294,7 @@ static inline uint8_t brModeAndFsToMaxSfbShort(const unsigned bitRateMode, const
return (samplingRate > 51200 ? 11 : 13) - 2 + (bitRateMode >> 2);
}
-#if SFB_QUANT_PERCEPT_OPT && !EE_MORE_MSE
+#if !EE_MORE_MSE
static inline void findActualBandwidthShort (uint8_t* const maxSfbShort, const uint16_t* sfbOffsets,
const int32_t* mdctSignals, const int32_t* mdstSignals, const unsigned nSamplesInShort)
{
@@ -322,7 +322,7 @@ static inline void findActualBandwidthShort (uint8_t* const maxSfbShort, const u
maxAbs = __max (maxAbs, abs (mdctSignals[e + w * nSamplesInShort]));
}
}
- if (maxAbs > maxSfb * (SA_EPS >> 1)) break;
+ if (maxAbs > maxSfb * (SA_EPS >> (3 - SFB_QUANT_PERCEPT_OPT))) break;
if (e == sfbOffs) sfbOffs = sfbOffsets[(--maxSfb) - 1];
}
@@ -761,9 +761,9 @@ uint32_t ExhaleEncoder::getThr (const unsigned channelIndex, const unsigned sfbI
uint32_t sumSfbLoud = 0;
for (int16_t s = 31; s >= 0; s--) sumSfbLoud += sfbLoudMem[s];
- sumSfbLoud = (sumSfbLoud + 32) >> 6; // -6 dB
+ sumSfbLoud = (sumSfbLoud + 16) >> 5;
- return sumSfbLoud * (sumSfbLoud >> (toSamplingRate (m_frequencyIdx) >> 13)); // scaled SMR
+ return (sumSfbLoud * sumSfbLoud + 3) >> 2;
}
unsigned ExhaleEncoder::psychBitAllocation () // perceptual bit-allocation via scale factors
@@ -776,7 +776,7 @@ unsigned ExhaleEncoder::psychBitAllocation () // perceptual bit-allocation via s
const uint8_t maxSfbLong = (useMaxBandwidth ? m_numSwbLong : brModeAndFsToMaxSfbLong (m_bitRateMode, samplingRate));
const uint16_t scaleSBR = (m_shiftValSBR > 0 || m_nonMpegExt ? sbrRateOffset[m_bitRateMode] : 0); // -25% rate
const uint64_t scaleSr = (samplingRate < 27713 ? (samplingRate < 23004 ? 32 : 34) - __min (3 << m_shiftValSBR, m_bitRateMode)
- : (m_bitRateMode != 3 && samplingRate < 37566 ? 36 : 37)) - (nChannels >> 1) - 4 + 4 * ((SFB_QUANT_PERCEPT_OPT + 1) / 2);
+ : (m_bitRateMode != 3 && samplingRate < 37566 ? 36 : 37) - 4 + 4 * ((SFB_QUANT_PERCEPT_OPT + 1) / 2)) - (nChannels >> 1);
const uint64_t scaleBr = (m_bitRateMode == 0 || m_frameCount <= 1 ? __min (32, 17u + (((samplingRate + (1 << 11)) >> 12) << 1) - (nChannels >> 1))
: scaleSr - eightTimesSqrt256Minus[256 - m_bitRateMode] - __min (3, (m_bitRateMode - 1) >> 1)) + scaleSBR;
uint32_t* sfbStepSizes = (uint32_t*) m_tempIntBuf;
@@ -975,7 +975,7 @@ unsigned ExhaleEncoder::psychBitAllocation () // perceptual bit-allocation via s
}
if (grpOff[maxSfbCh] > grpOff[0])
{
- s = BA_EPS + unsigned ((s * (eightShorts ? 1u + 55u/grpData.windowGroupLength[gr] : 7u) + 16383u) >> 14);
+ s = BA_EPS + unsigned ((s * (eightShorts ? 1u + 55u / grpData.windowGroupLength[gr] : 7u) + 16383u) >> 14);
# ifndef NO_PREROLL_DATA
if (((m_frameCount - 1u) % (m_indepPeriod << 1)) == 1 && m_numElements == 1 && !eightShorts) s = (4u + 9u * s) >> 3;
# endif
@@ -1013,6 +1013,52 @@ unsigned ExhaleEncoder::psychBitAllocation () // perceptual bit-allocation via s
#endif
grpScaleFacs[b] = m_bitAllocator.getScaleFac (grpStepSizes[b], &m_mdctSignals[ci][grpOff[b]], sfbWidth, grpRms[b]);
}
+#if !SFB_QUANT_PERCEPT_OPT
+ if (m_bitRateMode > 0 && m_numElements == 1)
+ {
+ if (maxSfbCh < (samplingRate < 18783 ? 17 : 24) || !m_frameCount)
+ {
+ for (s = 0; s < 26; s++) m_sfbLoudMem[ch][s][m_frameCount & 31] = uint16_t (m_frameCount ? sqrt ((double) getThr (ch, s)) : 32);
+ }
+ else // limit step-sizes around background noise level
+ {
+ const uint8_t* sd = coreConfig.stereoDataCurr;
+ const unsigned ns = __max (samplingRate < 27713 ? (samplingRate < 18783 ? 17 : 24) : 22, m_specGapFiller.getFirstGapFillSfb ());
+ uint32_t minFrRMS = TA_EPS >> EE_MORE_MSE;
+
+ for (s = ns; s < __min (ns + 26u, maxSfbCh); s++)
+ {
+ uint16_t* const lm = m_sfbLoudMem[ch][s - ns];
+ uint32_t minSfbMem = INT_MAX;
+
+ lm[m_frameCount & 31] = __max (BA_EPS, uint16_t (sqrt (double (sd[s] ? __max (grpRms[s], coreConfig.groupingData[1 - ch].sfbRmsValues[s]) : grpRms[s]))));
+
+ for (int f = 0; f < 32; f++)
+ {
+ const uint32_t slm = (uint32_t) lm[f] * lm[f];
+
+ if (minSfbMem > slm && slm > BA_EPS) minSfbMem = slm;
+ }
+ if (minSfbMem < INT_MAX)
+ {
+ const uint32_t w = grpOff[s + 1] - grpOff[s];
+
+ minSfbMem = (((minSfbMem + 2u) >> 2) + getThr (ch, s - ns) + w - 1u) / w;
+ if (minFrRMS > minSfbMem) minFrRMS = minSfbMem;
+ }
+ }
+ for (s -= ns; s < 26; s++) m_sfbLoudMem[ch][s][m_frameCount & 31] = BA_EPS;
+
+ for (s = (minFrRMS < (TA_EPS >> EE_MORE_MSE) ? ns : 99); s < maxSfbCh; s++)
+ {
+ const uint32_t w = grpOff[s + 1] - grpOff[s];
+
+ if (stepSizes[s] < minFrRMS * (w >> (sd[s] ? 1 : 0)))
+ grpScaleFacs[s] = m_bitAllocator.getScaleFac (minFrRMS * (w >> (sd[s] ? 1 : 0)), &m_mdctSignals[ci][grpOff[s]], w, grpRms[s]);
+ }
+ }
+ }
+#endif
} // for gr
#if !RESTRICT_TO_AAC
@@ -1026,7 +1072,7 @@ unsigned ExhaleEncoder::psychBitAllocation () // perceptual bit-allocation via s
if ((m_bitRateMode == 0) && (m_numElements == 1) && (samplingRate < 27713) && eightShorts)
{
- for (s = 0; s < 26; s++) m_sfbLoudMem[ch][s][m_frameCount & 31] = uint16_t (sqrt (double (getThr (ch, s) << (samplingRate >> 13))));
+ for (s = 0; s < 26; s++) m_sfbLoudMem[ch][s][m_frameCount & 31] = uint16_t (sqrt ((double) getThr (ch, s)));
}
if ((maxSfbCh < numSwbFrame) || (m_bitRateMode <= 2)) // increase coding bandwidth
{
@@ -1046,7 +1092,7 @@ unsigned ExhaleEncoder::psychBitAllocation () // perceptual bit-allocation via s
const unsigned sfbIdx = s - sfbStart;
m_sfbLoudMem[ch][sfbIdx][m_frameCount & 31] = __max (BA_EPS, uint16_t (sqrt (rmsValue)));
- if (grpRms[s] < getThr (ch, sfbIdx)) grpData.scaleFactors[s + m_numSwbShort * gr] = 0;
+ if (grpRms[s] < (getThr (ch, sfbIdx) >> (samplingRate >> 13))) grpData.scaleFactors[s + m_numSwbShort * gr] = 0;
}
}
else if ((m_bitRateMode <= 4) && (meanSpecFlat[ci] <= (SCHAR_MAX >> 1))) // lo
@@ -1424,7 +1470,7 @@ unsigned ExhaleEncoder::spectralProcessing () // complete ics_info(), calc TNS
m_specAnalyzer.getSpectralBandwidth (m_bandwidPrev, nChannels);
// spectral analysis for current MCLT signal (windowed time-samples for the current frame)
- errorValue |= m_specAnalyzer.spectralAnalysis (m_mdctSignals, m_mdstSignals, nChannels, nSamplesInFrame, samplingRate, lfeChannelIndex);
+ errorValue |= m_specAnalyzer.spectralAnalysis (m_mdctSignals, m_mdstSignals, nChannels, nSamplesInFrame, samplingRate, lfeChannelIndex, SFB_QUANT_PERCEPT_OPT);
// get spectral channel statistics for this frame, used for perceptual model & BW detector
m_specAnalyzer.getSpecAnalysisStats (m_specAnaCurr, nChannels);
@@ -1518,7 +1564,7 @@ unsigned ExhaleEncoder::spectralProcessing () // complete ics_info(), calc TNS
}
icsCurr.maxSfb = __min (icsCurr.maxSfb, brModeAndFsToMaxSfbLong (m_bitRateMode, samplingRate));
}
-#if SFB_QUANT_PERCEPT_OPT && !EE_MORE_MSE
+#if !EE_MORE_MSE
while (grpSO[icsCurr.maxSfb] > __max (m_bandwidCurr[ci], m_bandwidPrev[ci]) + (icsCurr.maxSfb >> 1)) icsCurr.maxSfb--; // detect BW
#endif
}
@@ -1551,7 +1597,7 @@ unsigned ExhaleEncoder::spectralProcessing () // complete ics_info(), calc TNS
}
}
memcpy (grpData.windowGroupLength, windowGroupingTable[icsCurr.windowGrouping], NUM_WINDOW_GROUPS * sizeof (uint8_t));
-#if SFB_QUANT_PERCEPT_OPT && !EE_MORE_MSE
+#if !EE_MORE_MSE
findActualBandwidthShort (&icsCurr.maxSfb, grpSO, m_mdctSignals[ci], nChannels < 2 ? nullptr : m_mdstSignals[ci], nSamplesInShort);
#endif
errorValue |= eightShortGrouping (grpData, grpSO, m_mdctSignals[ci], nChannels < 2 ? nullptr : m_mdstSignals[ci]);
@@ -2190,7 +2236,7 @@ unsigned ExhaleEncoder::initEncoder (unsigned char* const audioConfigBuffer, uin
m_elementData[el]->elementType = elementTypeConfig[chConf][el]; // usacElementType[el]
}
}
- memset (m_sfbLoudMem, 1, 2 * 26 * 32 * sizeof (uint16_t));
+ memset (m_sfbLoudMem, SFB_QUANT_PERCEPT_OPT, 2 * 26 * 32 * sizeof (uint16_t));
// allocate all signal buffers
if (m_shiftValSBR > 0)
diff --git a/src/lib/quantization.h b/src/lib/quantization.h
index 6c1af87..bb55b35 100644
--- a/src/lib/quantization.h
+++ b/src/lib/quantization.h
@@ -21,7 +21,7 @@
#define SF_THRESH_NEG 0.92044821 // round -1.5 dB
#define SF_THRESH_POS 1.09460356 // round +1.5 dB
#define SFB_QUANT_OFFSET 0.496094 // 13 - 29^(3/4)
-#define SFB_QUANT_PERCEPT_OPT 1 // psych. quant.
+#define SFB_QUANT_PERCEPT_OPT 2 // psych. quant.
#define QUANT_MAX 85 + (170 >> SFB_QUANT_PERCEPT_OPT)
// class for BL USAC quantization
diff --git a/src/lib/specAnalysis.cpp b/src/lib/specAnalysis.cpp
index 24f95cb..e48cbbc 100644
--- a/src/lib/specAnalysis.cpp
+++ b/src/lib/specAnalysis.cpp
@@ -1,11 +1,11 @@
/* specAnalysis.cpp - source file for class providing spectral analysis of MCLT signals
- * written by C. R. Helmrich, last modified in 2021 - see License.htm for legal notices
+ * written by C. R. Helmrich, last modified in 2025 - see License.htm for legal notices
*
* The copyright in this software is being made available under the exhale Copyright License
* and comes with ABSOLUTELY NO WARRANTY. This software may be subject to other third-
* party rights, including patent rights. No such rights are granted under this License.
*
- * Copyright (c) 2018-2021 Christian R. Helmrich, project ecodis. All rights reserved.
+ * Copyright (c) 2018-2025 Christian R. Helmrich, project ecodis. All rights reserved.
*/
#include "exhaleLibPch.h"
@@ -228,11 +228,11 @@ unsigned SpecAnalyzer::optimizeGrouping (const unsigned channelIndex, const unsi
unsigned SpecAnalyzer::spectralAnalysis (const int32_t* const mdctSignals[USAC_MAX_NUM_CHANNELS],
const int32_t* const mdstSignals[USAC_MAX_NUM_CHANNELS],
const unsigned nChannels, const unsigned nSamplesInFrame, const unsigned samplingRate,
- const unsigned lfeChannelIndex /*= USAC_MAX_NUM_CHANNELS*/) // to skip an LFE channel
+ const unsigned lfeChannelIndex /*= USAC_MAX_NUM_CHANNELS*/, const unsigned scale /*= 2*/)
{
const uint64_t anaBwOffset = SA_BW >> 1;
const unsigned lpcStopBand16k = (samplingRate <= 32000 ? nSamplesInFrame : (32000 * nSamplesInFrame) / samplingRate) >> SA_BW_SHIFT;
- const unsigned thresholdSlope = (48000 + SA_EPS * samplingRate) / 96000;
+ const unsigned thresholdSlope = (96000 + SA_EPS * scale * samplingRate) / 192000;
const unsigned thresholdStart = samplingRate >> 15;
if ((mdctSignals == nullptr) || (mdstSignals == nullptr) || (nChannels > USAC_MAX_NUM_CHANNELS) || (lfeChannelIndex > USAC_MAX_NUM_CHANNELS) ||
diff --git a/src/lib/specAnalysis.h b/src/lib/specAnalysis.h
index e7af357..79fb669 100644
--- a/src/lib/specAnalysis.h
+++ b/src/lib/specAnalysis.h
@@ -1,11 +1,11 @@
/* specAnalysis.h - header file for class providing spectral analysis of MCLT signals
- * written by C. R. Helmrich, last modified in 2021 - see License.htm for legal notices
+ * written by C. R. Helmrich, last modified in 2025 - see License.htm for legal notices
*
* The copyright in this software is being made available under the exhale Copyright License
* and comes with ABSOLUTELY NO WARRANTY. This software may be subject to other third-
* party rights, including patent rights. No such rights are granted under this License.
*
- * Copyright (c) 2018-2021 Christian R. Helmrich, project ecodis. All rights reserved.
+ * Copyright (c) 2018-2025 Christian R. Helmrich, project ecodis. All rights reserved.
*/
#ifndef _SPEC_ANALYSIS_H_
@@ -54,7 +54,7 @@ public:
unsigned spectralAnalysis (const int32_t* const mdctSignals[USAC_MAX_NUM_CHANNELS],
const int32_t* const mdstSignals[USAC_MAX_NUM_CHANNELS],
const unsigned nChannels, const unsigned nSamplesInFrame, const unsigned samplingRate,
- const unsigned lfeChannelIndex = USAC_MAX_NUM_CHANNELS); // to skip an LFE channel
+ const unsigned lfeChannelIndex = USAC_MAX_NUM_CHANNELS, const unsigned scale = 2);
int16_t stereoSigAnalysis (const int32_t* const mdctSignal1, const int32_t* const mdctSignal2,
const int32_t* const mdstSignal1, const int32_t* const mdstSignal2,
const unsigned nSamplesMax, const unsigned nSamplesInFrame, const bool shortTransforms,
|