opensmile-python
opensmile-python copied to clipboard
Published
20 hours ago •
audeering
audeering
How to adapt a opensmile config to pyopensmile?
I want to adapt the "embose2010" config to pyopensmile now. But I meet an error:
Traceback (most recent call last): File "D:\cocde\pratt\PraatScripts-master\opensimle_feature.py", line 12, in
smile = opensmile.Smile( File "D:\software\conda\lib\site-packages\opensmile\core\smile.py", line 116, in init self._feature_names(), File "D:\software\conda\lib\site-packages\opensmile\core\smile.py", line 275, in _feature_names smile = self._smile(options=options) File "D:\software\conda\lib\site-packages\opensmile\core\smile.py", line 366, in _smile smile.initialize( File "D:\software\conda\lib\site-packages\opensmile\core\SMILEapi.py", line 240, in initialize self._check_smile_result(smileapi.smile_initialize(self._smileobj, File "D:\software\conda\lib\site-packages\opensmile\core\SMILEapi.py", line 478, in _check_smile_result raise OpenSmileException(result) opensmile.core.SMILEapi.OpenSmileException: Code: 1
I only add a source and sink at the begining and end of the original config,like this:
///////////////////////////////////////////////////////////////////////////////////////
///////// > openSMILE configuration file for live emotion recognition < ///////////////
///////// base set of 988 features, 1st level functionals //////////////////
///////// of low-level descriptors such as MFCC, Pitch, LSP, ... //////////////////
///////// //////////////////
///////// * written 2009 by Florian Eyben * //////////////////
///////// //////////////////
///////// (c) 2014 audEERING UG (haftungsbeschränkt), //////////////////
///////// All rights reserverd. //////////////////
///////////////////////////////////////////////////////////////////////////////////////
// Modified version:
// - compatibility of newest 2.1 openSMILE code with openEAR 0.1.0 models
///////////////////////////////////////////////////////////////////////////////////////
;
; This section is always required in openSMILE configuration files
; it configures the componentManager and gives a list of all components which are to be loaded
; The order in which the components are listed should match
; the order of the data flow for most efficient processing
;
///////////////////////////////////////////////////////////////////////////////////////
[componentInstances:cComponentManager]
instance[dataMemory].type=cDataMemory
;;; default source
[componentInstances:cComponentManager]
instance[dataMemory].type=cDataMemory
;;; source
\{\cm[source{?}:include external source]}
;;; main section
[componentInstances:cComponentManager]
;this line configures the default data memory:
instance[dataMemory].type=cDataMemory
instance[waveIn].type=cWaveSource
instance[fr25].type=cFramer
;;; 40 ms frames features:
instance[fr40].type=cFramer
instance[w40].type=cWindower
instance[fft40].type=cTransformFFT
instance[fftmagphase40].type=cFFTmagphase
instance[acf40].type=cAcf
instance[cepstrum40].type=cAcf
; Pitch...
instance[pitchACF].type=cPitchACF
;;; 25 ms frames features:
instance[pe].type=cVectorPreemphasis
instance[win].type=cWindower
instance[fft].type=cTransformFFT
instance[fftmagphase].type=cFFTmagphase
instance[mspec].type=cMelspec
; MFCC
instance[mfcc].type=cMfcc
instance[lpc].type=cLpc
; Line Spectral Frequencies
instance[lsp].type=cLsp
; Zero-Crossings
instance[mzcr].type=cMZcr
; Intensity and Loudness (narrow-band approximation)
instance[intens].type=cIntensity
;;; all LLD concattenated and smoothed using a moving average filter
instance[lld].type=cContourSmoother
; delta coefficients of LLD
instance[delta1].type=cDeltaRegression
;;; functionals over FULL input (e.g. turns)
instance[functL1].type=cFunctionals
;;; write instances to a Weka ARFF file (comment out to disable)
instance[arffsink].type=cArffSink
;;; live classification of emotion (comment out lines to disable them):
;; run single threaded (nThreads=1)
; NOTE: a single thread is more efficient for processing small files, since multi-threaded processing involves more
; overhead during startup, which will make the system slower in the end
nThreads=1
;; do not show any internal dataMemory level settings
; (if you want to see them set the value to 1, 2, 3, or 4, depending on the amount of detail you wish)
printLevelStats=0
/////////////////////////////////////////////////////////////////////////////////////////////
///////////////////////// component configuration ////////////////////////////////////////
/////////////////////////////////////////////////////////////////////////////////////////////
; the following sections configure the components listed above
; a help on configuration parameters can be obtained with
; SMILExtract -H
; or
; SMILExtract -H configTypeName (= componentTypeName)
/////////////////////////////////////////////////////////////////////////////////////////////
[waveIn:cWaveSource]
; this sets the level this component writes to
; the level will be created by this component
; no other components may write to a level having the same name
writer.dmLevel=wave
; this defines a new commandline option "-I" or "-inputfile", which can be used to specify
; the filename on the commandline instead of having it "hard-coded" in the config file
filename=\cm[inputfile(I){test.wav}:name of input file]
; mix stereo files down to mono for analysis
monoMixdown = 1
properTimestamps = 1
start = 0.0
end = -1
[fr40:cFramer]
reader.dmLevel=wave
writer.dmLevel=frames40
frameSize = 0.040
frameStep = 0.010
frameMode = fixed
frameCenterSpecial = center
[w40:cWindower]
reader.dmLevel=frames40
writer.dmLevel=win40frame
copyInputName = 1
processArrayFields = 1
winFunc = ham
gain = 1.0
offset = 0
[fft40:cTransformFFT]
reader.dmLevel=win40frame
writer.dmLevel=fftc40
copyInputName = 1
processArrayFields = 1
inverse = 0
; for compatibility with 2.2.0 and older versions
zeroPadSymmetric = 0
[fftmagphase40:cFFTmagphase]
reader.dmLevel=fftc40
writer.dmLevel=fftmag40
copyInputName = 1
processArrayFields = 1
inverse = 0
magnitude = 1
phase = 0
[acf40:cAcf]
reader.dmLevel=fftmag40
writer.dmLevel=acf40
nameAppend = acf
copyInputName = 1
processArrayFields = 1
usePower = 1
cepstrum = 0
acfCepsNormOutput = 0
[cepstrum40:cAcf]
reader.dmLevel=fftmag40
writer.dmLevel=cepstrum40
nameAppend = acf
copyInputName = 1
processArrayFields = 1
usePower = 1
cepstrum = 1
acfCepsNormOutput = 0
oldCompatCepstrum = 1
absCepstrum = 1
[pitchACF:cPitchACF]
; the pitchACF component must ALWAYS read from acf AND cepstrum in the given order!
reader.dmLevel=acf40;cepstrum40
writer.dmLevel=pitch
processArrayFields=0
maxPitch = 500
voiceProb = 1
voiceQual = 0
HNR = 0
F0 = 1
F0raw = 0
F0env = 1
voicingCutoff = 0.550000
[fr25:cFramer]
reader.dmLevel=wave
writer.dmLevel=frames
frameSize = 0.025
frameStep = 0.010
frameMode = fixed
frameCenterSpecial = center
[pe:cVectorPreemphasis]
reader.dmLevel=frames
writer.dmLevel=framespe
copyInputName = 1
processArrayFields = 1
k=0.97
[win:cWindower]
reader.dmLevel=framespe
writer.dmLevel=winframe
copyInputName = 1
processArrayFields = 1
winFunc = ham
gain = 1.0
offset = 0
[fft:cTransformFFT]
reader.dmLevel=winframe
writer.dmLevel=fftc
copyInputName = 1
processArrayFields = 1
inverse = 0
; for compatibility with 2.2.0 and older versions
zeroPadSymmetric = 0
[fftmagphase:cFFTmagphase]
reader.dmLevel=fftc
writer.dmLevel=fftmag
copyInputName = 1
processArrayFields = 1
inverse = 0
magnitude = 1
phase = 0
[mspec:cMelspec]
reader.dmLevel=fftmag
writer.dmLevel=mspec1
copyInputName = 1
processArrayFields = 1
htkcompatible = 1
nBands = 26
lofreq = 0
hifreq = 8000
usePower = 1
inverse = 0
specScale = mel
[mfcc:cMfcc]
reader.dmLevel=mspec1
writer.dmLevel=mfcc1
copyInputName = 0
processArrayFields = 1
firstMfcc = 1
lastMfcc = 12
cepLifter = 22.0
htkcompatible = 1
[lpc:cLpc]
reader.dmLevel=framespe
writer.dmLevel=lpc
copyInputName = 1
processArrayFields = 1
method = acf
p = 8
saveLPCoeff = 1
lpGain = 0
saveRefCoeff = 0
residual = 0
forwardFilter = 0
lpSpectrum = 0
[lsp:cLsp]
reader.dmLevel=lpc
writer.dmLevel=lsp
copyInputName = 1
processArrayFields = 0
[intens:cIntensity]
reader.dmLevel=frames
writer.dmLevel=intens
copyInputName = 1
processArrayFields = 1
intensity=1
loudness =1
[mzcr:cMZcr]
reader.dmLevel=frames
writer.dmLevel=mzcr
copyInputName = 1
processArrayFields = 1
zcr = 1
amax = 0
mcr = 0
maxmin = 0
dc = 0
[lld:cContourSmoother]
reader.dmLevel=intens;mfcc1;lsp;mzcr;pitch
writer.dmLevel=lld
writer.levelconf.nT = 2500
writer.levelconf.isRb=0
writer.levelconf.growDyn=1
nameAppend = sma
copyInputName = 1
noPostEOIprocessing = 0
smaWin = 3
// ---- delta regression of LLD ----
[delta1:cDeltaRegression]
reader.dmLevel=lld
writer.dmLevel=lld_de
blocksize=1
writer.levelconf.nT = 2500
writer.levelconf.isRb=0
writer.levelconf.growDyn=1
nameAppend = de
copyInputName = 1
noPostEOIprocessing = 0
deltawin=2
// statistical functionals
[functL1:cFunctionals]
reader.dmLevel=lld;lld_de
writer.dmLevel=func
copyInputName = 1
; frameMode = var will enable the functionals component to listen for messages from the turn detector
frameMode = full
functionalsEnabled=Extremes;Regression;Moments;Percentiles
Extremes.max = 1
Extremes.min = 1
Extremes.range = 1
Extremes.maxpos = 1
Extremes.minpos = 1
Extremes.amean = 1
Extremes.maxameandist = 0
Extremes.minameandist = 0
; Note: the much better way to normalise the times of maxpos and minpos
; is 'turn', however for compatibility with old files the default 'frame'
; is kept here:
Extremes.norm = frame
Regression.linregc1 = 1
Regression.linregc2 = 1
Regression.linregerrA = 1
Regression.linregerrQ = 1
Regression.qregc1 = 0
Regression.qregc2 = 0
Regression.qregc3 = 0
Regression.qregerrA = 0
Regression.qregerrQ = 0
Regression.centroid = 0
Regression.doRatioLimit = 0
Regression.centroidRatioLimit = 0
Moments.doRatioLimit = 0
Moments.variance = 0
Moments.stddev = 1
Moments.skewness = 1
Moments.kurtosis = 1
Moments.amean = 0
Percentiles.quartiles = 1
Percentiles.iqr = 1
//////////////////////////////////////////////////////////////////////
/////////////////// data output configuration //////////////////////
//////////////////////////////////////////////////////////////////////
// ----- you can use this to save the features extracted ------
[arffsink:cArffSink]
reader.dmLevel=func
; do not print "frameNumber" attribute to ARFF file
number=0
; name of output file as commandline option
filename=\cm[arffout(O){output.arff}:name of WEKA Arff output file]
; name of @relation in the ARFF file
relation=\cm[corpus{SMILEfeaturesLive}:corpus name, arff relation]
; base name of the current instance, turn number will be appended
instanceBase=liveturn
; name of class label
class[0].name = emotion
; list of nominal classes OR "numeric"
class[0].type = \cm[classes{unknown}:all classes for arff file attribute]
; the class label or value for the current instance
target[0].all = \cm[classlabel{unassigned}:instance class label]
; ** NOTE: theoretically the classified class label could be assigned here, however this would require
; ** saving the class label along with the frame in the dataMemory
; ** or sending this meta-data via a message (including frame number)
; ** neither solution is currently implemented.... :-(
;
append=0
//////---------------------- END -------------------------///////
;;; sink
\{\cm[sink{?}:include external sink]}
What else should I do? Thanks a lot !
You still have the old cWaveSource in your script. Here is the fixed script:
///////////////////////////////////////////////////////////////////////////////////////
///////// > openSMILE configuration file for live emotion recognition < ///////////////
///////// base set of 988 features, 1st level functionals //////////////////
///////// of low-level descriptors such as MFCC, Pitch, LSP, ... //////////////////
///////// //////////////////
///////// * written 2009 by Florian Eyben * //////////////////
///////// //////////////////
///////// (c) 2014 audEERING UG (haftungsbeschränkt), //////////////////
///////// All rights reserverd. //////////////////
///////////////////////////////////////////////////////////////////////////////////////
// Modified version:
// - compatibility of newest 2.1 openSMILE code with openEAR 0.1.0 models
///////////////////////////////////////////////////////////////////////////////////////
;
; This section is always required in openSMILE configuration files
; it configures the componentManager and gives a list of all components which are to be loaded
; The order in which the components are listed should match
; the order of the data flow for most efficient processing
;
///////////////////////////////////////////////////////////////////////////////////////
[componentInstances:cComponentManager]
instance[dataMemory].type=cDataMemory
;;; default source
[componentInstances:cComponentManager]
instance[dataMemory].type=cDataMemory
;;; source
\{\cm[source{?}:include external source]}
;;; main section
[componentInstances:cComponentManager]
;this line configures the default data memory:
instance[dataMemory].type=cDataMemory
instance[fr25].type=cFramer
;;; 40 ms frames features:
instance[fr40].type=cFramer
instance[w40].type=cWindower
instance[fft40].type=cTransformFFT
instance[fftmagphase40].type=cFFTmagphase
instance[acf40].type=cAcf
instance[cepstrum40].type=cAcf
; Pitch...
instance[pitchACF].type=cPitchACF
;;; 25 ms frames features:
instance[pe].type=cVectorPreemphasis
instance[win].type=cWindower
instance[fft].type=cTransformFFT
instance[fftmagphase].type=cFFTmagphase
instance[mspec].type=cMelspec
; MFCC
instance[mfcc].type=cMfcc
instance[lpc].type=cLpc
; Line Spectral Frequencies
instance[lsp].type=cLsp
; Zero-Crossings
instance[mzcr].type=cMZcr
; Intensity and Loudness (narrow-band approximation)
instance[intens].type=cIntensity
;;; all LLD concattenated and smoothed using a moving average filter
instance[lld].type=cContourSmoother
; delta coefficients of LLD
instance[delta1].type=cDeltaRegression
;;; functionals over FULL input (e.g. turns)
instance[functL1].type=cFunctionals
;;; write instances to a Weka ARFF file (comment out to disable)
instance[arffsink].type=cArffSink
;;; live classification of emotion (comment out lines to disable them):
;; run single threaded (nThreads=1)
; NOTE: a single thread is more efficient for processing small files, since multi-threaded processing involves more
; overhead during startup, which will make the system slower in the end
nThreads=1
;; do not show any internal dataMemory level settings
; (if you want to see them set the value to 1, 2, 3, or 4, depending on the amount of detail you wish)
printLevelStats=0
/////////////////////////////////////////////////////////////////////////////////////////////
///////////////////////// component configuration ////////////////////////////////////////
/////////////////////////////////////////////////////////////////////////////////////////////
; the following sections configure the components listed above
; a help on configuration parameters can be obtained with
; SMILExtract -H
; or
; SMILExtract -H configTypeName (= componentTypeName)
/////////////////////////////////////////////////////////////////////////////////////////////
[fr40:cFramer]
reader.dmLevel=wave
writer.dmLevel=frames40
frameSize = 0.040
frameStep = 0.010
frameMode = fixed
frameCenterSpecial = center
[w40:cWindower]
reader.dmLevel=frames40
writer.dmLevel=win40frame
copyInputName = 1
processArrayFields = 1
winFunc = ham
gain = 1.0
offset = 0
[fft40:cTransformFFT]
reader.dmLevel=win40frame
writer.dmLevel=fftc40
copyInputName = 1
processArrayFields = 1
inverse = 0
; for compatibility with 2.2.0 and older versions
zeroPadSymmetric = 0
[fftmagphase40:cFFTmagphase]
reader.dmLevel=fftc40
writer.dmLevel=fftmag40
copyInputName = 1
processArrayFields = 1
inverse = 0
magnitude = 1
phase = 0
[acf40:cAcf]
reader.dmLevel=fftmag40
writer.dmLevel=acf40
nameAppend = acf
copyInputName = 1
processArrayFields = 1
usePower = 1
cepstrum = 0
acfCepsNormOutput = 0
[cepstrum40:cAcf]
reader.dmLevel=fftmag40
writer.dmLevel=cepstrum40
nameAppend = acf
copyInputName = 1
processArrayFields = 1
usePower = 1
cepstrum = 1
acfCepsNormOutput = 0
oldCompatCepstrum = 1
absCepstrum = 1
[pitchACF:cPitchACF]
; the pitchACF component must ALWAYS read from acf AND cepstrum in the given order!
reader.dmLevel=acf40;cepstrum40
writer.dmLevel=pitch
processArrayFields=0
maxPitch = 500
voiceProb = 1
voiceQual = 0
HNR = 0
F0 = 1
F0raw = 0
F0env = 1
voicingCutoff = 0.550000
[fr25:cFramer]
reader.dmLevel=wave
writer.dmLevel=frames
frameSize = 0.025
frameStep = 0.010
frameMode = fixed
frameCenterSpecial = center
[pe:cVectorPreemphasis]
reader.dmLevel=frames
writer.dmLevel=framespe
copyInputName = 1
processArrayFields = 1
k=0.97
[win:cWindower]
reader.dmLevel=framespe
writer.dmLevel=winframe
copyInputName = 1
processArrayFields = 1
winFunc = ham
gain = 1.0
offset = 0
[fft:cTransformFFT]
reader.dmLevel=winframe
writer.dmLevel=fftc
copyInputName = 1
processArrayFields = 1
inverse = 0
; for compatibility with 2.2.0 and older versions
zeroPadSymmetric = 0
[fftmagphase:cFFTmagphase]
reader.dmLevel=fftc
writer.dmLevel=fftmag
copyInputName = 1
processArrayFields = 1
inverse = 0
magnitude = 1
phase = 0
[mspec:cMelspec]
reader.dmLevel=fftmag
writer.dmLevel=mspec1
copyInputName = 1
processArrayFields = 1
htkcompatible = 1
nBands = 26
lofreq = 0
hifreq = 8000
usePower = 1
inverse = 0
specScale = mel
[mfcc:cMfcc]
reader.dmLevel=mspec1
writer.dmLevel=mfcc1
copyInputName = 0
processArrayFields = 1
firstMfcc = 1
lastMfcc = 12
cepLifter = 22.0
htkcompatible = 1
[lpc:cLpc]
reader.dmLevel=framespe
writer.dmLevel=lpc
copyInputName = 1
processArrayFields = 1
method = acf
p = 8
saveLPCoeff = 1
lpGain = 0
saveRefCoeff = 0
residual = 0
forwardFilter = 0
lpSpectrum = 0
[lsp:cLsp]
reader.dmLevel=lpc
writer.dmLevel=lsp
copyInputName = 1
processArrayFields = 0
[intens:cIntensity]
reader.dmLevel=frames
writer.dmLevel=intens
copyInputName = 1
processArrayFields = 1
intensity=1
loudness =1
[mzcr:cMZcr]
reader.dmLevel=frames
writer.dmLevel=mzcr
copyInputName = 1
processArrayFields = 1
zcr = 1
amax = 0
mcr = 0
maxmin = 0
dc = 0
[lld:cContourSmoother]
reader.dmLevel=intens;mfcc1;lsp;mzcr;pitch
writer.dmLevel=lld
writer.levelconf.nT = 2500
writer.levelconf.isRb=0
writer.levelconf.growDyn=1
nameAppend = sma
copyInputName = 1
noPostEOIprocessing = 0
smaWin = 3
// ---- delta regression of LLD ----
[delta1:cDeltaRegression]
reader.dmLevel=lld
writer.dmLevel=lld_de
blocksize=1
writer.levelconf.nT = 2500
writer.levelconf.isRb=0
writer.levelconf.growDyn=1
nameAppend = de
copyInputName = 1
noPostEOIprocessing = 0
deltawin=2
// statistical functionals
[functL1:cFunctionals]
reader.dmLevel=lld;lld_de
writer.dmLevel=func
copyInputName = 1
; frameMode = var will enable the functionals component to listen for messages from the turn detector
frameMode = full
functionalsEnabled=Extremes;Regression;Moments;Percentiles
Extremes.max = 1
Extremes.min = 1
Extremes.range = 1
Extremes.maxpos = 1
Extremes.minpos = 1
Extremes.amean = 1
Extremes.maxameandist = 0
Extremes.minameandist = 0
; Note: the much better way to normalise the times of maxpos and minpos
; is 'turn', however for compatibility with old files the default 'frame'
; is kept here:
Extremes.norm = frame
Regression.linregc1 = 1
Regression.linregc2 = 1
Regression.linregerrA = 1
Regression.linregerrQ = 1
Regression.qregc1 = 0
Regression.qregc2 = 0
Regression.qregc3 = 0
Regression.qregerrA = 0
Regression.qregerrQ = 0
Regression.centroid = 0
Regression.doRatioLimit = 0
Regression.centroidRatioLimit = 0
Moments.doRatioLimit = 0
Moments.variance = 0
Moments.stddev = 1
Moments.skewness = 1
Moments.kurtosis = 1
Moments.amean = 0
Percentiles.quartiles = 1
Percentiles.iqr = 1
//////////////////////////////////////////////////////////////////////
/////////////////// data output configuration //////////////////////
//////////////////////////////////////////////////////////////////////
// ----- you can use this to save the features extracted ------
[arffsink:cArffSink]
reader.dmLevel=func
; do not print "frameNumber" attribute to ARFF file
number=0
; name of output file as commandline option
filename=\cm[arffout(O){output.arff}:name of WEKA Arff output file]
; name of @relation in the ARFF file
relation=\cm[corpus{SMILEfeaturesLive}:corpus name, arff relation]
; base name of the current instance, turn number will be appended
instanceBase=liveturn
; name of class label
class[0].name = emotion
; list of nominal classes OR "numeric"
class[0].type = \cm[classes{unknown}:all classes for arff file attribute]
; the class label or value for the current instance
target[0].all = \cm[classlabel{unassigned}:instance class label]
; ** NOTE: theoretically the classified class label could be assigned here, however this would require
; ** saving the class label along with the frame in the dataMemory
; ** or sending this meta-data via a message (including frame number)
; ** neither solution is currently implemented.... :-(
;
append=0
//////---------------------- END -------------------------///////
;;; sink
\{\cm[sink{?}:include external sink]}
Btw: when you get opensmile.core.SMILEapi.OpenSmileException: Code: 1, it's always a good idea to create a log file using the logfile argument. For your script it would print:
[ 17.02.2021 - 11:14:34 ]
(MSG) [2] SMILEapi: openSMILE starting!
[ 17.02.2021 - 11:14:34 ]
(MSG) [2] SMILEapi: config file is: /media/jwagner/Data/Git/pysmile-public/my/test.conf
[ 17.02.2021 - 11:14:34 ]
(MSG) [2] cComponentManager: successfully registered 103 component types.
[ 17.02.2021 - 11:14:34 ]
(ERR) [0] commandlineParser: parse: unknown option '-bufferModeRbConf' on commandline!
[ 17.02.2021 - 11:14:34 ]
(ERR) [0] commandlineParser: parse: unknown option '-frameModeFunctionalsConf' on commandline!
[ 17.02.2021 - 11:14:34 ]
(ERR) [1] dataMemory: two components cannot write to the same level: 'wave', component1='extsource.writer', component2='waveIn.writer'
You still have the old
cWaveSourcein your script. Here is the fixed script://///////////////////////////////////////////////////////////////////////////////////// ///////// > openSMILE configuration file for live emotion recognition < /////////////// ///////// base set of 988 features, 1st level functionals ////////////////// ///////// of low-level descriptors such as MFCC, Pitch, LSP, ... ////////////////// ///////// ////////////////// ///////// * written 2009 by Florian Eyben * ////////////////// ///////// ////////////////// ///////// (c) 2014 audEERING UG (haftungsbeschränkt), ////////////////// ///////// All rights reserverd. ////////////////// /////////////////////////////////////////////////////////////////////////////////////// // Modified version: // - compatibility of newest 2.1 openSMILE code with openEAR 0.1.0 models /////////////////////////////////////////////////////////////////////////////////////// ; ; This section is always required in openSMILE configuration files ; it configures the componentManager and gives a list of all components which are to be loaded ; The order in which the components are listed should match ; the order of the data flow for most efficient processing ; /////////////////////////////////////////////////////////////////////////////////////// [componentInstances:cComponentManager] instance[dataMemory].type=cDataMemory ;;; default source [componentInstances:cComponentManager] instance[dataMemory].type=cDataMemory ;;; source \{\cm[source{?}:include external source]} ;;; main section [componentInstances:cComponentManager] ;this line configures the default data memory: instance[dataMemory].type=cDataMemory instance[fr25].type=cFramer ;;; 40 ms frames features: instance[fr40].type=cFramer instance[w40].type=cWindower instance[fft40].type=cTransformFFT instance[fftmagphase40].type=cFFTmagphase instance[acf40].type=cAcf instance[cepstrum40].type=cAcf ; Pitch... instance[pitchACF].type=cPitchACF ;;; 25 ms frames features: instance[pe].type=cVectorPreemphasis instance[win].type=cWindower instance[fft].type=cTransformFFT instance[fftmagphase].type=cFFTmagphase instance[mspec].type=cMelspec ; MFCC instance[mfcc].type=cMfcc instance[lpc].type=cLpc ; Line Spectral Frequencies instance[lsp].type=cLsp ; Zero-Crossings instance[mzcr].type=cMZcr ; Intensity and Loudness (narrow-band approximation) instance[intens].type=cIntensity ;;; all LLD concattenated and smoothed using a moving average filter instance[lld].type=cContourSmoother ; delta coefficients of LLD instance[delta1].type=cDeltaRegression ;;; functionals over FULL input (e.g. turns) instance[functL1].type=cFunctionals ;;; write instances to a Weka ARFF file (comment out to disable) instance[arffsink].type=cArffSink ;;; live classification of emotion (comment out lines to disable them): ;; run single threaded (nThreads=1) ; NOTE: a single thread is more efficient for processing small files, since multi-threaded processing involves more ; overhead during startup, which will make the system slower in the end nThreads=1 ;; do not show any internal dataMemory level settings ; (if you want to see them set the value to 1, 2, 3, or 4, depending on the amount of detail you wish) printLevelStats=0 ///////////////////////////////////////////////////////////////////////////////////////////// ///////////////////////// component configuration //////////////////////////////////////// ///////////////////////////////////////////////////////////////////////////////////////////// ; the following sections configure the components listed above ; a help on configuration parameters can be obtained with ; SMILExtract -H ; or ; SMILExtract -H configTypeName (= componentTypeName) ///////////////////////////////////////////////////////////////////////////////////////////// [fr40:cFramer] reader.dmLevel=wave writer.dmLevel=frames40 frameSize = 0.040 frameStep = 0.010 frameMode = fixed frameCenterSpecial = center [w40:cWindower] reader.dmLevel=frames40 writer.dmLevel=win40frame copyInputName = 1 processArrayFields = 1 winFunc = ham gain = 1.0 offset = 0 [fft40:cTransformFFT] reader.dmLevel=win40frame writer.dmLevel=fftc40 copyInputName = 1 processArrayFields = 1 inverse = 0 ; for compatibility with 2.2.0 and older versions zeroPadSymmetric = 0 [fftmagphase40:cFFTmagphase] reader.dmLevel=fftc40 writer.dmLevel=fftmag40 copyInputName = 1 processArrayFields = 1 inverse = 0 magnitude = 1 phase = 0 [acf40:cAcf] reader.dmLevel=fftmag40 writer.dmLevel=acf40 nameAppend = acf copyInputName = 1 processArrayFields = 1 usePower = 1 cepstrum = 0 acfCepsNormOutput = 0 [cepstrum40:cAcf] reader.dmLevel=fftmag40 writer.dmLevel=cepstrum40 nameAppend = acf copyInputName = 1 processArrayFields = 1 usePower = 1 cepstrum = 1 acfCepsNormOutput = 0 oldCompatCepstrum = 1 absCepstrum = 1 [pitchACF:cPitchACF] ; the pitchACF component must ALWAYS read from acf AND cepstrum in the given order! reader.dmLevel=acf40;cepstrum40 writer.dmLevel=pitch processArrayFields=0 maxPitch = 500 voiceProb = 1 voiceQual = 0 HNR = 0 F0 = 1 F0raw = 0 F0env = 1 voicingCutoff = 0.550000 [fr25:cFramer] reader.dmLevel=wave writer.dmLevel=frames frameSize = 0.025 frameStep = 0.010 frameMode = fixed frameCenterSpecial = center [pe:cVectorPreemphasis] reader.dmLevel=frames writer.dmLevel=framespe copyInputName = 1 processArrayFields = 1 k=0.97 [win:cWindower] reader.dmLevel=framespe writer.dmLevel=winframe copyInputName = 1 processArrayFields = 1 winFunc = ham gain = 1.0 offset = 0 [fft:cTransformFFT] reader.dmLevel=winframe writer.dmLevel=fftc copyInputName = 1 processArrayFields = 1 inverse = 0 ; for compatibility with 2.2.0 and older versions zeroPadSymmetric = 0 [fftmagphase:cFFTmagphase] reader.dmLevel=fftc writer.dmLevel=fftmag copyInputName = 1 processArrayFields = 1 inverse = 0 magnitude = 1 phase = 0 [mspec:cMelspec] reader.dmLevel=fftmag writer.dmLevel=mspec1 copyInputName = 1 processArrayFields = 1 htkcompatible = 1 nBands = 26 lofreq = 0 hifreq = 8000 usePower = 1 inverse = 0 specScale = mel [mfcc:cMfcc] reader.dmLevel=mspec1 writer.dmLevel=mfcc1 copyInputName = 0 processArrayFields = 1 firstMfcc = 1 lastMfcc = 12 cepLifter = 22.0 htkcompatible = 1 [lpc:cLpc] reader.dmLevel=framespe writer.dmLevel=lpc copyInputName = 1 processArrayFields = 1 method = acf p = 8 saveLPCoeff = 1 lpGain = 0 saveRefCoeff = 0 residual = 0 forwardFilter = 0 lpSpectrum = 0 [lsp:cLsp] reader.dmLevel=lpc writer.dmLevel=lsp copyInputName = 1 processArrayFields = 0 [intens:cIntensity] reader.dmLevel=frames writer.dmLevel=intens copyInputName = 1 processArrayFields = 1 intensity=1 loudness =1 [mzcr:cMZcr] reader.dmLevel=frames writer.dmLevel=mzcr copyInputName = 1 processArrayFields = 1 zcr = 1 amax = 0 mcr = 0 maxmin = 0 dc = 0 [lld:cContourSmoother] reader.dmLevel=intens;mfcc1;lsp;mzcr;pitch writer.dmLevel=lld writer.levelconf.nT = 2500 writer.levelconf.isRb=0 writer.levelconf.growDyn=1 nameAppend = sma copyInputName = 1 noPostEOIprocessing = 0 smaWin = 3 // ---- delta regression of LLD ---- [delta1:cDeltaRegression] reader.dmLevel=lld writer.dmLevel=lld_de blocksize=1 writer.levelconf.nT = 2500 writer.levelconf.isRb=0 writer.levelconf.growDyn=1 nameAppend = de copyInputName = 1 noPostEOIprocessing = 0 deltawin=2 // statistical functionals [functL1:cFunctionals] reader.dmLevel=lld;lld_de writer.dmLevel=func copyInputName = 1 ; frameMode = var will enable the functionals component to listen for messages from the turn detector frameMode = full functionalsEnabled=Extremes;Regression;Moments;Percentiles Extremes.max = 1 Extremes.min = 1 Extremes.range = 1 Extremes.maxpos = 1 Extremes.minpos = 1 Extremes.amean = 1 Extremes.maxameandist = 0 Extremes.minameandist = 0 ; Note: the much better way to normalise the times of maxpos and minpos ; is 'turn', however for compatibility with old files the default 'frame' ; is kept here: Extremes.norm = frame Regression.linregc1 = 1 Regression.linregc2 = 1 Regression.linregerrA = 1 Regression.linregerrQ = 1 Regression.qregc1 = 0 Regression.qregc2 = 0 Regression.qregc3 = 0 Regression.qregerrA = 0 Regression.qregerrQ = 0 Regression.centroid = 0 Regression.doRatioLimit = 0 Regression.centroidRatioLimit = 0 Moments.doRatioLimit = 0 Moments.variance = 0 Moments.stddev = 1 Moments.skewness = 1 Moments.kurtosis = 1 Moments.amean = 0 Percentiles.quartiles = 1 Percentiles.iqr = 1 ////////////////////////////////////////////////////////////////////// /////////////////// data output configuration ////////////////////// ////////////////////////////////////////////////////////////////////// // ----- you can use this to save the features extracted ------ [arffsink:cArffSink] reader.dmLevel=func ; do not print "frameNumber" attribute to ARFF file number=0 ; name of output file as commandline option filename=\cm[arffout(O){output.arff}:name of WEKA Arff output file] ; name of @relation in the ARFF file relation=\cm[corpus{SMILEfeaturesLive}:corpus name, arff relation] ; base name of the current instance, turn number will be appended instanceBase=liveturn ; name of class label class[0].name = emotion ; list of nominal classes OR "numeric" class[0].type = \cm[classes{unknown}:all classes for arff file attribute] ; the class label or value for the current instance target[0].all = \cm[classlabel{unassigned}:instance class label] ; ** NOTE: theoretically the classified class label could be assigned here, however this would require ; ** saving the class label along with the frame in the dataMemory ; ** or sending this meta-data via a message (including frame number) ; ** neither solution is currently implemented.... :-( ; append=0 //////---------------------- END -------------------------/////// ;;; sink \{\cm[sink{?}:include external sink]}
Thanks！ It really help me ！！
Is there a simple way to adapt any of the config files compatible woth the executable version to the Python version?
In most cases replacing:
[waveIn:cWaveSource]
writer.dmLevel=wave
with
\{\cm[source{?}:include external source]}
and adding
\{\cm[sink{?}:include external sink]}
should work.