malar
/
speech-scoring
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

merged

master
Malar Kannan 2017-11-28 17:05:20 +05:30
parent f12da988d3
commit ec7303223c
1 changed files with 22 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

60
speech_pitch.py
View File

@ -5,36 +5,35 @@ import matplotlib.pyplot as plt
import seaborn as sns import seaborn as sns
sns.set() # Use seaborn's default style to make graphs more pretty sns.set() # Use seaborn's default style to make graphs more pretty
def pitch_array(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'):
def pm_snd(sample_file):
# sample_file = 'inputs/self-apple/apple-low1.aiff'
samples, samplerate, _ = snd.read(sample_file) samples, samplerate, _ = snd.read(sample_file)
sample_sound = pm.Sound(values=samples,sampling_frequency=samplerate) return pm.Sound(values=samples,sampling_frequency=samplerate)
def pitch_array(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'):
sample_sound = pm_snd(sample_file)
sample_pitch = sample_sound.to_pitch() sample_pitch = sample_sound.to_pitch()
return sample_pitch.to_matrix().as_array() return sample_pitch.to_matrix().as_array()
def intensity_array(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'): def intensity_array(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'):
sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff' sample_sound = pm_snd(sample_file)
samples, samplerate, _ = snd.read(sample_file)
sample_sound = pm.Sound(values=samples,sampling_frequency=samplerate)
sample_intensity = sample_sound.to_mfcc() sample_intensity = sample_sound.to_mfcc()
sample_intensity.as_array().shape sample_intensity.as_array().shape
return sample_pitch.to_matrix().as_array() return sample_pitch.to_matrix().as_array()
def compute_mfcc(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'): def compute_mfcc(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'):
# sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff' sample_sound = pm_snd(sample_file)
samples, samplerate, _ = snd.read(sample_file)
sample_sound = pm.Sound(values=samples,sampling_frequency=samplerate)
sample_mfcc = sample_sound.to_mfcc() sample_mfcc = sample_sound.to_mfcc()
# sample_mfcc.to_array().shape # sample_mfcc.to_array().shape
return sample_mfcc.to_array() return sample_mfcc.to_array()
def compute_formants(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'): def compute_formants(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'):
sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff' sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'
samples, samplerate, _ = snd.read(sample_file) sample_sound = pm_snd(sample_file)
sample_sound = pm.Sound(values=samples,sampling_frequency=samplerate)
sample_formant = sample_sound.to_formant_burg() sample_formant = sample_sound.to_formant_burg()
sample_formant.x_bins() # sample_formant.x_bins()
# sample_mfcc.to_array().shape return sample_formant.x_bins()
return sample_mfcc.to_array()
def draw_spectrogram(spectrogram, dynamic_range=70): def draw_spectrogram(spectrogram, dynamic_range=70):
X, Y = spectrogram.x_grid(), spectrogram.y_grid() X, Y = spectrogram.x_grid(), spectrogram.y_grid()
@ -62,10 +61,16 @@ def draw_pitch(pitch):
plt.ylim(0, pitch.ceiling) plt.ylim(0, pitch.ceiling)
plt.ylabel("pitch [Hz]") plt.ylabel("pitch [Hz]")
def pm_snd(sample_file): def draw_formants(formant):
# sample_file = 'inputs/self-apple/apple-low1.aiff' # Extract selected pitch contour, and
samples, samplerate, _ = snd.read(sample_file) # replace unvoiced samples by NaN to not plot
return pm.Sound(values=samples,sampling_frequency=samplerate) formant_values = formant.to_matrix().values
pitch_values[pitch_values==0] = np.nan
plt.plot(pitch.xs(), pitch_values, linewidth=3, color='w')
plt.plot(pitch.xs(), pitch_values, linewidth=1)
plt.grid(False)
plt.ylim(0, pitch.ceiling)
plt.ylabel("Formants [val]")
def plot_sample_raw(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'): def plot_sample_raw(sample_file='outputs/audio/sunflowers-Victoria-180-normal-870.aiff'):
@ -111,29 +116,8 @@ def plot_sample_pitch(sample_file='outputs/audio/sunflowers-Victoria-180-normal-
if __name__ == '__main__': if __name__ == '__main__':
# sunflowers_vic_180_norm = pitch_array('outputs/audio/sunflowers-Victoria-180-normal-870.aiff')
# sunflowers_fred_180_norm = pitch_array('outputs/audio/sunflowers-Fred-180-normal-6515.aiff')
# sunflowers_vic_180_norm_mfcc = compute_mfcc('outputs/audio/sunflowers-Victoria-180-normal-870.aiff')
# fred_180_norm_mfcc = compute_mfcc('outputs/audio/sunflowers-Fred-180-normal-6515.aiff')
# alex_mfcc = compute_mfcc('outputs/audio/sunflowers-Alex-180-normal-4763.aiff')
# # # sunflowers_vic_180_norm.shape
# # # sunflowers_fred_180_norm.shape
# # alex_mfcc.shape
# # sunflowers_vic_180_norm_mfcc.shape
# # sunflowers_fred_180_norm_mfcc.shape
# from speech_spectrum import generate_aiff_spectrogram
# vic_spec = generate_aiff_spectrogram('outputs/audio/sunflowers-Victoria-180-normal-870.aiff')
# alex_spec = generate_aiff_spectrogram('outputs/audio/sunflowers-Alex-180-normal-4763.aiff')
# alex150spec = generate_aiff_spectrogram('outputs/audio/sunflowers-Alex-150-normal-589.aiff')
# vic_spec.shape
# alex_spec.shape
# alex150spec.shape
# alex_mfcc.shape
# fred_180_norm_mfcc.shape
mom_snd = pm_snd('outputs/test/moms_are_engineers-7608.aiff') mom_snd = pm_snd('outputs/test/moms_are_engineers-7608.aiff')
plot_sample_pitch('outputs/audio/sunflowers-Victoria-180-normal-870.aiff') plot_sample_pitch('outputs/audio/sunflowers-Victoria-180-normal-870.aiff')
plot_sample_pitch('inputs/self-apple/apple-low1.aiff') plot_sample_pitch('inputs/self-apple/apple-low1.aiff')
plot_sample_pitch('inputs/self-apple/apple-low2.aiff') plot_sample_pitch('inputs/self-apple/apple-low2.aiff')
plot_sample_pitch('inputs/self-apple/apple-medium1.aiff') plot_sample_pitch('inputs/self-apple/apple-medium1.aiff')
# pm.SoundFileFormat
# pm.Pitch.get_number_of_frames()