nkululeko 0.59.1__py3-none-any.whl → 0.61.0__py3-none-any.whl

nkululeko/feinberg_praat.py

@@ -1,430 +0,0 @@
-"""
-This is a copy of David R. Feinberg's Praat scripts
-https://github.com/drfeinberg/PraatScripts
-taken June 23rd 2022
-"""
-
-#!/usr/bin/env python3
-import numpy as np
-import pandas as pd
-import math
-import parselmouth 
-import statistics
-from nkululeko.util import Util
-import audiofile
-from parselmouth.praat import call
-from scipy.stats.mstats import zscore
-from sklearn.decomposition import PCA
-from sklearn.preprocessing import StandardScaler
-
-
-# This is the function to measure source acoustics using default male parameters.
-
-def measurePitch(voiceID, f0min, f0max, unit):
-    sound = parselmouth.Sound(voiceID) # read the sound
-    duration = call(sound, "Get total duration") # duration
-    pitch = call(sound, "To Pitch", 0.0, f0min, f0max) #create a praat pitch object
-    meanF0 = call(pitch, "Get mean", 0, 0, unit) # get mean pitch
-    stdevF0 = call(pitch, "Get standard deviation", 0 ,0, unit) # get standard deviation
-    harmonicity = call(sound, "To Harmonicity (cc)", 0.01, f0min, 0.1, 1.0)
-    hnr = call(harmonicity, "Get mean", 0, 0)
-    pointProcess = call(sound, "To PointProcess (periodic, cc)", f0min, f0max)
-    localJitter = call(pointProcess, "Get jitter (local)", 0, 0, 0.0001, 0.02, 1.3)
-    localabsoluteJitter = call(pointProcess, "Get jitter (local, absolute)", 0, 0, 0.0001, 0.02, 1.3)
-    rapJitter = call(pointProcess, "Get jitter (rap)", 0, 0, 0.0001, 0.02, 1.3)
-    ppq5Jitter = call(pointProcess, "Get jitter (ppq5)", 0, 0, 0.0001, 0.02, 1.3)
-    ddpJitter = call(pointProcess, "Get jitter (ddp)", 0, 0, 0.0001, 0.02, 1.3)
-    localShimmer =  call([sound, pointProcess], "Get shimmer (local)", 0, 0, 0.0001, 0.02, 1.3, 1.6)
-    localdbShimmer = call([sound, pointProcess], "Get shimmer (local_dB)", 0, 0, 0.0001, 0.02, 1.3, 1.6)
-    apq3Shimmer = call([sound, pointProcess], "Get shimmer (apq3)", 0, 0, 0.0001, 0.02, 1.3, 1.6)
-    aqpq5Shimmer = call([sound, pointProcess], "Get shimmer (apq5)", 0, 0, 0.0001, 0.02, 1.3, 1.6)
-    apq11Shimmer =  call([sound, pointProcess], "Get shimmer (apq11)", 0, 0, 0.0001, 0.02, 1.3, 1.6)
-    ddaShimmer = call([sound, pointProcess], "Get shimmer (dda)", 0, 0, 0.0001, 0.02, 1.3, 1.6)
-    
-    return duration, meanF0, stdevF0, hnr, localJitter, localabsoluteJitter, rapJitter, ppq5Jitter, ddpJitter, localShimmer, localdbShimmer, apq3Shimmer, aqpq5Shimmer, apq11Shimmer, ddaShimmer
-
-
-# ## This function measures formants at each glottal pulse
-# 
-# Puts, D. A., Apicella, C. L., & Cárdenas, R. A. (2012). Masculine voices signal men's threat potential in forager and industrial societies. Proceedings of the Royal Society of London B: Biological Sciences, 279(1728), 601-609.
-# 
-# Adapted from: DOI 10.17605/OSF.IO/K2BHS
-# This function measures formants using Formant Position formula
-#def measureFormants(sound, wave_file, f0min,f0max):
-def measureFormants(sound, f0min,f0max):
-    sound = parselmouth.Sound(sound) # read the sound
-#    pitch = call(sound, "To Pitch (cc)", 0, f0min, 15, 'no', 0.03, 0.45, 0.01, 0.35, 0.14, f0max)
-    pointProcess = call(sound, "To PointProcess (periodic, cc)", f0min, f0max)
-    
-    formants = call(sound, "To Formant (burg)", 0.0025, 5, 5000, 0.025, 50)
-    numPoints = call(pointProcess, "Get number of points")
-
-    f1_list = []
-    f2_list = []
-    f3_list = []
-    f4_list = []
-    
-    # Measure formants only at glottal pulses
-    for point in range(0, numPoints):
-        point += 1
-        t = call(pointProcess, "Get time from index", point)
-        f1 = call(formants, "Get value at time", 1, t, 'Hertz', 'Linear')
-        f2 = call(formants, "Get value at time", 2, t, 'Hertz', 'Linear')
-        f3 = call(formants, "Get value at time", 3, t, 'Hertz', 'Linear')
-        f4 = call(formants, "Get value at time", 4, t, 'Hertz', 'Linear')
-        f1_list.append(f1)
-        f2_list.append(f2)
-        f3_list.append(f3)
-        f4_list.append(f4)
-    
-    f1_list = [f1 for f1 in f1_list if str(f1) != 'nan']
-    f2_list = [f2 for f2 in f2_list if str(f2) != 'nan']
-    f3_list = [f3 for f3 in f3_list if str(f3) != 'nan']
-    f4_list = [f4 for f4 in f4_list if str(f4) != 'nan']
-    
-    # calculate mean formants across pulses
-    f1_mean = statistics.mean(f1_list)
-    f2_mean = statistics.mean(f2_list)
-    f3_mean = statistics.mean(f3_list)
-    f4_mean = statistics.mean(f4_list)
-    
-    # calculate median formants across pulses, this is what is used in all subsequent calcualtions
-    # you can use mean if you want, just edit the code in the boxes below to replace median with mean
-    f1_median = statistics.median(f1_list)
-    f2_median = statistics.median(f2_list)
-    f3_median = statistics.median(f3_list)
-    f4_median = statistics.median(f4_list)
-    
-    return f1_mean, f2_mean, f3_mean, f4_mean, f1_median, f2_median, f3_median, f4_median
-
-
-# ## This function runs a 2-factor Principle Components Analysis (PCA) on Jitter and Shimmer
-
-def runPCA(df):
-    # z-score the Jitter and Shimmer measurements
-    measures = ['localJitter', 'localabsoluteJitter', 'rapJitter', 'ppq5Jitter', 'ddpJitter',
-                'localShimmer', 'localdbShimmer', 'apq3Shimmer', 'apq5Shimmer', 'apq11Shimmer', 'ddaShimmer']
-    x = df.loc[:, measures].values
-    # f = open('x.pickle', 'wb')
-    # pickle.dump(x, f)
-    # f.close()
-
-    x = StandardScaler().fit_transform(x)
-    if np.any(np.isnan(x)):
-        print (f'Warning: {np.count_nonzero(np.isnan(x))} Nans in x, replacing with 0')
-        x[np.isnan(x)] = 0
-    if np.any(np.isfinite(x)):
-        print (f'Warning: {np.count_nonzero(np.isfinite(x))} infinite in x')
-    
-    # PCA
-    pca = PCA(n_components=2)
-    principalComponents = pca.fit_transform(x)
-    if np.any(np.isnan(principalComponents)):
-        print ('pc is nan')
-        print(f'count: {np.count_nonzero(np.isnan(principalComponents))}')
-        print(principalComponents)
-        principalComponents=np.nan_to_num(principalComponents)
-
-    principalDf = pd.DataFrame(data = principalComponents, columns = ['JitterPCA', 'ShimmerPCA'])
-
-    return principalDf
-
-
-# ## This block of code runs the above functions on all of the '.wav' files in the /audio folder
-
-def compute_features(file_index):
-    # create lists to put the results
-    file_list = []
-    duration_list = []
-    mean_F0_list = []
-    sd_F0_list = []
-    hnr_list = []
-    localJitter_list = []
-    localabsoluteJitter_list = []
-    rapJitter_list = []
-    ppq5Jitter_list = []
-    ddpJitter_list = []
-    localShimmer_list = []
-    localdbShimmer_list = []
-    apq3Shimmer_list = []
-    aqpq5Shimmer_list = []
-    apq11Shimmer_list = []
-    ddaShimmer_list = []
-    f1_mean_list = []
-    f2_mean_list = []
-    f3_mean_list = []
-    f4_mean_list = []
-    f1_median_list = []
-    f2_median_list = []
-    f3_median_list = []
-    f4_median_list = []
-    # Go through all the wave files in the folder and measure all the acoustics
-#    for i, wave_file in enumerate(file_list):
-    for idx, (wave_file, start, end) in enumerate(file_index.to_list()):
-        signal, sampling_rate = audiofile.read(wave_file, offset=start.total_seconds(), duration=(end-start).total_seconds(), always_2d=True)
-        sound = parselmouth.Sound(values=signal, sampling_frequency=sampling_rate)
-        if idx%10==0:
-            print(f'praat: extracting file {idx} of {len(file_index.to_list())}')
-        #sound = parselmouth.Sound(wave_file)
-        (duration, meanF0, stdevF0, hnr, localJitter, localabsoluteJitter, rapJitter, ppq5Jitter, ddpJitter, 
-        localShimmer, localdbShimmer, apq3Shimmer, aqpq5Shimmer, apq11Shimmer, ddaShimmer) = measurePitch(
-            sound, 75, 300, "Hertz")
-        (f1_mean, f2_mean, f3_mean, f4_mean, f1_median, f2_median, f3_median, f4_median) = measureFormants(
-            sound, 75, 300)
-#        file_list.append(wave_file) # make an ID list
-        duration_list.append(duration) # make duration list
-        mean_F0_list.append(meanF0) # make a mean F0 list
-        sd_F0_list.append(stdevF0) # make a sd F0 list
-        hnr_list.append(hnr) #add HNR data
-        
-        # add raw jitter and shimmer measures
-        localJitter_list.append(localJitter)
-        localabsoluteJitter_list.append(localabsoluteJitter)
-        rapJitter_list.append(rapJitter)
-        ppq5Jitter_list.append(ppq5Jitter)
-        ddpJitter_list.append(ddpJitter)
-        localShimmer_list.append(localShimmer)
-        localdbShimmer_list.append(localdbShimmer)
-        apq3Shimmer_list.append(apq3Shimmer)
-        aqpq5Shimmer_list.append(aqpq5Shimmer)
-        apq11Shimmer_list.append(apq11Shimmer)
-        ddaShimmer_list.append(ddaShimmer)
-        
-        # add the formant data
-        f1_mean_list.append(f1_mean)
-        f2_mean_list.append(f2_mean)
-        f3_mean_list.append(f3_mean)
-        f4_mean_list.append(f4_mean)
-        f1_median_list.append(f1_median)
-        f2_median_list.append(f2_median)
-        f3_median_list.append(f3_median)
-        f4_median_list.append(f4_median)
-
-    # ## This block of code adds all of that data we just generated to a Pandas data frame
-    # Add the data to Pandas
-    df = pd.DataFrame(np.column_stack([duration_list, mean_F0_list, sd_F0_list, hnr_list, 
-                                    localJitter_list, localabsoluteJitter_list, rapJitter_list, 
-                                    ppq5Jitter_list, ddpJitter_list, localShimmer_list, 
-                                    localdbShimmer_list, apq3Shimmer_list, aqpq5Shimmer_list, 
-                                    apq11Shimmer_list, ddaShimmer_list, f1_mean_list, 
-                                    f2_mean_list, f3_mean_list, f4_mean_list, 
-                                    f1_median_list, f2_median_list, f3_median_list, 
-                                    f4_median_list]),
-                                    columns=['duration', 'meanF0Hz', 'stdevF0Hz', 'HNR', 
-                                                'localJitter', 'localabsoluteJitter', 'rapJitter', 
-                                                'ppq5Jitter', 'ddpJitter', 'localShimmer', 
-                                                'localdbShimmer', 'apq3Shimmer', 'apq5Shimmer', 
-                                                'apq11Shimmer', 'ddaShimmer', 'f1_mean', 'f2_mean', 
-                                                'f3_mean', 'f4_mean', 'f1_median', 
-                                                'f2_median', 'f3_median', 'f4_median'])
-
-    # add pca data
-    pcaData = runPCA(df) # Run jitter and shimmer PCA
-    df = pd.concat([df, pcaData], axis=1) # Add PCA data
-    # reload the data so it's all numbers
-    df.to_csv("processed_results.csv", index=False)
-    df = pd.read_csv('processed_results.csv', header=0)
-#    df.sort_values('voiceID').head(20)
-    # ## Next we calculate the vocal-tract length estimates
-
-    # ### Formant position
-    #  Puts, D. A., Apicella, C. L., & Cárdenas, R. A. (2012). Masculine voices signal men's threat potential in forager and industrial societies. Proceedings of the Royal Society of London B: Biological Sciences, 279(1728), 601-609.
-
-    df['pF'] = (zscore(df.f1_median) + zscore(df.f2_median) + zscore(df.f3_median) + zscore(df.f4_median)) / 4
-
-    # ### Formant Dispersion
-    # Fitch, W. T. (1997). Vocal tract length and formant frequency dispersion correlate with body size in rhesus macaques. The Journal of the Acoustical Society of America, 102(2), 1213-1222.
-
-
-    df['fdisp'] = (df['f4_median'] - df['f1_median']) / 3
-
-
-    # ### Fn (Average Formant)
-    # Pisanski, K., & Rendall, D. (2011). The prioritization of voice fundamental frequency or formants in listeners’ assessments of speaker size, masculinity, and attractiveness. The Journal of the Acoustical Society of America, 129(4), 2201-2212.
-
-    df['avgFormant'] = (df['f1_median'] + df['f2_median'] + df['f3_median'] + df['f4_median']) / 4
-
-    # ### MFF 
-    # Smith, D. R., & Patterson, R. D. (2005). The interaction of glottal-pulse rate and vocal-tract length in judgements of speaker size, sex, and age. The Journal of the Acoustical Society of America, 118(5), 3177-3186.
-
-    df['mff'] = (df['f1_median'] * df['f2_median'] * df['f3_median'] * df['f4_median']) ** 0.25
-
-
-    # ### Fitch VTL
-    # Fitch, W. T. (1997). Vocal tract length and formant frequency dispersion correlate with body size in rhesus macaques. The Journal of the Acoustical Society of America, 102(2), 1213-1222.
-
-
-    # reload the data again
-    #df.to_csv("processed_results.csv", index=False)
-    #df = pd.read_csv('processed_results.csv', header=0)
-
-    df['fitch_vtl'] = ((1 * (35000 / (4 * df['f1_median']))) +
-                    (3 * (35000 / (4 * df['f2_median']))) + 
-                    (5 * (35000 / (4 * df['f3_median']))) + 
-                    (7 * (35000 / (4 * df['f4_median'])))) / 4
-
-
-    # ### $\Delta$F 
-    # Reby,D.,& McComb,K.(2003). Anatomical constraints generate honesty: acoustic cues to age and weight in the roars of red deer stags. Animal Behaviour, 65, 519e-530.
-
-
-    xysum = (0.5 * df['f1_median']) + (1.5 * df['f2_median']) + (2.5 * df['f3_median']) + (3.5 * df['f4_median'])
-    xsquaredsum = (0.5 ** 2) + (1.5 ** 2) + (2.5 ** 2) + (3.5 ** 2)
-    df['delta_f'] = xysum / xsquaredsum
-
-
-    # ### VTL($\Delta$F)
-    # Reby,D.,&McComb,K.(2003).Anatomical constraints generate honesty: acoustic cues to age and weight in the roars of red deer stags. Animal Behaviour, 65, 519e-530.
-
-
-    df['vtl_delta_f'] = 35000 / (2 * df['delta_f'])
-    
-    print('Now extracting speech rate parameters...')
-
-    df_speechrate = get_speech_rate(file_index)
-    print('')
-    
-    return df.join(df_speechrate)
-
-
-
-
-"""
-Speech rate script taken from https://github.com/drfeinberg/PraatScripts
-on 25/05/23
-"""
-
-def get_speech_rate(file_index):
-    cols = ['nsyll', 'npause', 'dur(s)', 'phonationtime(s)', 'speechrate(nsyll / dur)', 'articulation '
-            'rate(nsyll / phonationtime)', 'ASD(speakingtime / nsyll)']
-    datalist = []
-    for idx, (wave_file, start, end) in enumerate(file_index.to_list()):
-        signal, sampling_rate = audiofile.read(wave_file, offset=start.total_seconds(), duration=(end-start).total_seconds(), always_2d=True)
-        sound = parselmouth.Sound(values=signal, sampling_frequency=sampling_rate)
-        # print(f'processing {file}')
-        speechrate_dictionary = speech_rate(sound)
-        datalist.append(speechrate_dictionary)
-        if idx%10==0:
-            print('.', end=' ')
-    df = pd.DataFrame(datalist)
-    return df 
-
-def speech_rate(sound):
-    silencedb = -25
-    mindip = 2
-    minpause = 0.3
-    originaldur = sound.get_total_duration()
-    intensity = sound.to_intensity(50)
-    start = call(intensity, "Get time from frame number", 1)
-    nframes = call(intensity, "Get number of frames")
-    end = call(intensity, "Get time from frame number", nframes)
-    min_intensity = call(intensity, "Get minimum", 0, 0, "Parabolic")
-    max_intensity = call(intensity, "Get maximum", 0, 0, "Parabolic")
-
-    # get .99 quantile to get maximum (without influence of non-speech sound bursts)
-    max_99_intensity = call(intensity, "Get quantile", 0, 0, 0.99)
-
-    # estimate Intensity threshold
-    threshold = max_99_intensity + silencedb
-    threshold2 = max_intensity - max_99_intensity
-    threshold3 = silencedb - threshold2
-    if threshold < min_intensity:
-        threshold = min_intensity
-
-    # get pauses (silences) and speakingtime
-    textgrid = call(intensity, "To TextGrid (silences)", threshold3, minpause, 0.1, "silent", "sounding")
-    silencetier = call(textgrid, "Extract tier", 1)
-    silencetable = call(silencetier, "Down to TableOfReal", "sounding")
-    npauses = call(silencetable, "Get number of rows")
-    speakingtot = 0
-    for ipause in range(npauses):
-        pause = ipause + 1
-        beginsound = call(silencetable, "Get value", pause, 1)
-        endsound = call(silencetable, "Get value", pause, 2)
-        speakingdur = endsound - beginsound
-        speakingtot += speakingdur
-
-    intensity_matrix = call(intensity, "Down to Matrix")
-    # sndintid = sound_from_intensity_matrix
-    sound_from_intensity_matrix = call(intensity_matrix, "To Sound (slice)", 1)
-    # use total duration, not end time, to find out duration of intdur (intensity_duration)
-    # in order to allow nonzero starting times.
-    intensity_duration = call(sound_from_intensity_matrix, "Get total duration")
-    intensity_max = call(sound_from_intensity_matrix, "Get maximum", 0, 0, "Parabolic")
-    point_process = call(sound_from_intensity_matrix, "To PointProcess (extrema)", "Left", "yes", "no", "Sinc70")
-    # estimate peak positions (all peaks)
-    numpeaks = call(point_process, "Get number of points")
-    t = [call(point_process, "Get time from index", i + 1) for i in range(numpeaks)]
-
-    # fill array with intensity values
-    timepeaks = []
-    peakcount = 0
-    intensities = []
-    for i in range(numpeaks):
-        value = call(sound_from_intensity_matrix, "Get value at time", t[i], "Cubic")
-        if value > threshold:
-            peakcount += 1
-            intensities.append(value)
-            timepeaks.append(t[i])
-
-    # fill array with valid peaks: only intensity values if preceding
-    # dip in intensity is greater than mindip
-    validpeakcount = 0
-    currenttime = timepeaks[0]
-    currentint = intensities[0]
-    validtime = []
-
-    for p in range(peakcount - 1):
-        following = p + 1
-        followingtime = timepeaks[p + 1]
-        dip = call(intensity, "Get minimum", currenttime, timepeaks[p + 1], "None")
-        diffint = abs(currentint - dip)
-        if diffint > mindip:
-            validpeakcount += 1
-            validtime.append(timepeaks[p])
-        currenttime = timepeaks[following]
-        currentint = call(intensity, "Get value at time", timepeaks[following], "Cubic")
-
-    # Look for only voiced parts
-    pitch = sound.to_pitch_ac(0.02, 30, 4, False, 0.03, 0.25, 0.01, 0.35, 0.25, 450)
-    voicedcount = 0
-    voicedpeak = []
-
-    for time in range(validpeakcount):
-        querytime = validtime[time]
-        whichinterval = call(textgrid, "Get interval at time", 1, querytime)
-        whichlabel = call(textgrid, "Get label of interval", 1, whichinterval)
-        value = pitch.get_value_at_time(querytime) 
-        if not math.isnan(value):
-            if whichlabel == "sounding":
-                voicedcount += 1
-                voicedpeak.append(validtime[time])
-
-    # calculate time correction due to shift in time for Sound object versus
-    # intensity object
-    timecorrection = originaldur / intensity_duration
-
-    # Insert voiced peaks in TextGrid
-    call(textgrid, "Insert point tier", 1, "syllables")
-    for i in range(len(voicedpeak)):
-        position = (voicedpeak[i] * timecorrection)
-        call(textgrid, "Insert point", 1, position, "")
-
-    # return results
-    speakingrate = voicedcount / originaldur
-    articulationrate = voicedcount / speakingtot
-    npause = npauses - 1
-    try:
-        asd = speakingtot / voicedcount
-    except ZeroDivisionError:
-        asd = 0
-        print('caught zero division')
-    speechrate_dictionary = {'nsyll':voicedcount,
-                             'npause': npause,
-                             'dur(s)':originaldur,
-                             'phonationtime(s)':intensity_duration,
-                             'speechrate(nsyll / dur)': speakingrate,
-                             "articulation rate(nsyll / phonationtime)":articulationrate,
-                             "ASD(speakingtime / nsyll)":asd}
-    return speechrate_dictionary
-

nkululeko/loss_ccc.py

@@ -1,28 +0,0 @@
-# loss_concordance_cor_coeff.py
-
-import torch 
-
-class ConcordanceCorCoeff(torch.nn.Module):
-
-    def __init__(self):
-        super().__init__()
-        self.mean = torch.mean
-        self.var = torch.var
-        self.sum = torch.sum
-        self.sqrt = torch.sqrt
-        self.std = torch.std
-
-    def forward(self, prediction, ground_truth):
-        mean_gt = self.mean(ground_truth, 0)
-        mean_pred = self.mean(prediction, 0)
-        var_gt = self.var (ground_truth, 0)
-        var_pred = self.var (prediction, 0)
-        v_pred = prediction - mean_pred
-        v_gt = ground_truth - mean_gt
-        cor = self.sum (v_pred * v_gt) / (self.sqrt(self.sum(v_pred ** 2)) * self.sqrt(self.sum(v_gt ** 2)))
-        sd_gt = self.std(ground_truth)
-        sd_pred = self.std(prediction)
-        numerator=2*cor*sd_gt*sd_pred
-        denominator=var_gt+var_pred+(mean_gt-mean_pred)**2
-        ccc = numerator/denominator
-        return 1-ccc

nkululeko/loss_softf1loss.py

@@ -1,40 +0,0 @@
-import torch
-from torch.nn.functional import one_hot, softmax
-
-class SoftF1Loss(torch.nn.Module):
-
-    ''' differentiable F1 loss, adapted from
-    https://gist.github.com/SuperShinyEyes/dcc68a08ff8b615442e3bc6a9b55a354 
-    
-    written by Uwe Reichel'''
-
-    def __init__(self, weight=None, epsilon=1e-7, num_classes=-1):
-        super().__init__()
-        self.weight = weight
-        self.epsilon = epsilon
-        self.num_classes = num_classes
-
-    def forward(self, y_pred, y_true):
-        assert y_pred.ndim == 2
-        assert y_true.ndim == 1
-        
-        y_true = one_hot(y_true, self.num_classes).to(torch.float32)
-        y_pred = softmax(y_pred, dim=1)
-
-        tp = (y_true * y_pred).sum(dim=0).to(torch.float32)
-        fp = ((1 - y_true) * y_pred).sum(dim=0).to(torch.float32)
-        fn = (y_true * (1 - y_pred)).sum(dim=0).to(torch.float32)
-
-        precision = tp / (tp + fp + self.epsilon)
-        recall = tp / (tp + fn + self.epsilon)
-
-        f1 = 2 * (precision * recall) / (precision + recall + self.epsilon)
-        f1 = f1.clamp(min=self.epsilon, max=1-self.epsilon)
-
-        # unweighted mean F1 loss
-        if self.weight is None:
-            return 1 - f1.mean()
-
-        # weighted mean F1 loss
-        wm = torch.sum(f1 * self.weight) / torch.sum(self.weight)        
-        return 1 - wm