API - Documentation¶
mlend.download_spoken_numerals¶
Help on function download_spoken_numerals in module mlend.downloader:
- download_spoken_numerals(save_to=’../MLEnd’, subset={}, verbose=1, overwrite=False, pbar_style=’colab’)
Download Spoken Numerals Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/spoken_numerals/).
- subset: subset of data to download. {‘attribute_name’:list_of_values to select}
subset = {‘Numeral’:[1,2]}, will download audio files of numerals 1 and 2 only subset = {‘Numeral’:[1,2], ‘Intonation’:[‘excited’,’question’], ‘Speaker’:[1,2,3,4,5,6]} subset = {} will download entire dataset
- # Raises
- ValueError:
in case keys in subset are not in [‘Numeral’, ‘Intonation’, ‘Speaker’]
in case values of any keys are not valid
# Returns
path: path where data is saved
mlend.download_hums_whistles¶
Help on function download_hums_whistles in module mlend.downloader:
- download_hums_whistles(save_to=’../MLEnd’, subset={}, verbose=1, overwrite=False, pbar_style=’colab’)
Download Hums and Whistles Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/hums_whistles/).
- subset: subset of data to download. {‘attribute_name’:list_of_values to select}
subset = {‘Song’:[‘Potter’,’Frozen’]}, will download audio files of Potter and Fronzen only with both Hums and Whistles subset = {‘Song’:[‘Potter’,’Frozen’], ‘Interpretation’:[‘Hum’]} will download audio files of Potter and Fronzen with Huming only subset = {‘Interpretation’:[‘Whistle’]} will download all the audio files of Whistling subset = {} will download entire dataset
- # Raises
- ValueError:
in case keys in subset are not in [‘Song’, ‘Interpretation’, ‘Interpreter’]
in case values of any keys are not valid
# Returns
path: path where data is saved
mlend.download_london_sounds¶
Help on function download_london_sounds in module mlend.downloader:
- download_london_sounds(save_to=’../MLEnd’, subset={}, verbose=1, overwrite=False, pbar_style=’colab’)
Download London Sounds Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/london_sounds/).
- subset: subset of data to download. {‘attribute_name’:list_of_values to select}
subset = {‘Area’:[‘british_museum’], ‘Spot’:[‘forecourt’,’greatcourt’]}, will download audio files of British museum for two spots only ‘forecourt’,’greatcourt’ subset = {‘Area’:[‘british_museum’]} will download all the spots of British Museum subset = {‘Area’:[‘british_museum’],’In_Out’:[‘indoor’]} will download all the indoor spots of British Museum subset = {} will download entire dataset
- # Raises
- ValueError:
in case keys in subset are not in [‘Area’, ‘Spot’, ‘In_Out’]
in case values of any keys are not valid
# Returns
path: path where data is saved
mlend.download_yummy_small¶
Help on function download_yummy_small in module mlend.downloader:
- download_yummy_small(save_to=’../MLEnd’, verbose=1, overwrite=False, pbar_style=’colab’)
Download Yummy Small Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/yummy/).
# Returns
path: path where data is saved
mlend.download_yummy¶
Help on function download_yummy in module mlend.downloader:
- download_yummy(save_to=’../MLEnd’, subset={}, verbose=1, overwrite=False, pbar_style=’colab’, debug_mode=False)
Download Yummy Full Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/yummy/).
- subset: subset of data to download. {‘attribute_name’:list_of_values to select}
subset = {‘Diet’:[‘vegetarian’]}, will download image files of vegetarian dishes only subset = {‘Diet’:[‘vegan’], ‘Home_or_restaurant’:[‘home’]} will download image files of vegan dishes cooked at home subset = {} will download entire dataset
- # Raises
- ValueError:
in case keys in subset are not in the attribuate list [‘Diet’, ‘Home_or_restaurant’, ‘Cuisine’ ]
in case values of any keys are not valid
# Returns
path: path where data is saved
mlend.download_happiness¶
Help on function download_happiness in module mlend.downloader:
- download_happiness(save_to=’../MLEnd’, verbose=1, overwrite=False)
Download Happiness Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/happiness/).
# Returns
path: path where data is saved
mlend.download_load_happiness¶
Help on function download_load_happiness in module mlend.downloader:
- download_load_happiness()
Download Happiness Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/spoken_numerals/).
# Returns
pandas dataframe
mlend.spoken_numerals_load¶
Help on function spoken_numerals_load in module mlend.processing:
- spoken_numerals_load(datadir_main=’../MLEnd/spoken_numerals’, train_test_split=’Benchmark_A’, verbose=1, encode_labels=True)
Read files of Spoken Numerals Dataset and create training and testing sets.
- # Arguments
- datadir_main (str): local path where ‘MLEndSND_audiofiles’ directory is stored
relative to ../MLEnd/spoken_numerals/).
- train_test_split (str): split type for training and testing
- ‘Benchmark_A’: Speaker Independent Benchmark
Training (70%) and Testing (30%) do not have any common speaker
- ‘Benchmark_B’: Speaker Dependent Benchmark
Training (70%) and Testing (30%) both sets have same speakers
‘Random’ or ‘random’: random split woth 70-30
float (e.g. 0.8) (>0 and <1) random split with given fraction for training set. if train_test_split = 0.8, Training set will be 80% and Testing 20%
encode_labels: (bool), if to encode labels
- # Raises
- ValueError:
if train_test_split is not str [‘Benchmark_A’, ‘Benchmark_B’, ‘random’] or float (<1 and >0)”
- # Returns
TrainSet: A dictionary with keys {‘X_paths’, ‘Y’, ‘Y_encoded’} TestSet: A dictionary with keys {‘X_paths’, ‘Y’, ‘Y_encoded’}
‘X_paths’ is list of paths for audio files
‘Y’ is Nx3 np.array, column 0 for Numerals, 1 for Intonation and 2 for Speaker
- ‘Y_encoded’ is Nx3 np.array same as ‘Y’, column 0 for Numerals, 1 for Intonation and 2 for Speaker
each column is encoded as 0, 1, 2 ..
MAPs : A dictionary of maps, if encode_labels is true, else an empty dictionary
mlend.hums_whistles_load¶
Help on function hums_whistles_load in module mlend.processing:
- hums_whistles_load(datadir_main=’../MLEnd/hums_whistles’, train_test_split=’Benchmark_A’, verbose=1, encode_labels=True)
Read files of Hums and Whistles Dataset and create training and testing sets.
- # Arguments
- datadir_main (str): local path where ‘MLEndHWD_audiofiles’ directory is stored
relative to ../MLEnd/hums_whistles/).
- train_test_split (str): split type for training and testing
- ‘Benchmark_A’: Speaker Independent Benchmark
Training (70%) and Testing (30%) do not have any common speaker
- ‘Benchmark_B’: Speaker Dependent Benchmark
Training (70%) and Testing (30%) both sets have same speakers
‘Random’ or ‘random’: random split woth 70-30
float (e.g. 0.8) (>0 and <1) random split with given fraction for training set. if train_test_split = 0.8, Training set will be 80% and Testing 20%
encode_labels: (bool), if to encode labels
- # Raises
- ValueError:
if train_test_split is not str [‘Benchmark_A’, ‘Benchmark_B’, ‘random’] or float (<1 and >0)”
- # Returns
TrainSet: A dictionary with keys {‘X_paths’, ‘Y’, ‘Y_encoded’} TestSet: A dictionary with keys {‘X_paths’, ‘Y’, ‘Y_encoded’}
‘X_paths’ is list of paths for audio files
‘Y’ is Nx3 np.array, column 0 for Song, 1 for Interpretation, and 2 for Interpreter
- ‘Y_encoded’ is Nx3 np.array same as ‘Y’, column 0 for Song, 1 for Interpretation and 2 for Interpreter
each column is encoded as 0, 1, 2 ..
MAPs : A dictionary of maps, if encode_labels is true, else an empty dictionary
mlend.london_sounds_load¶
Help on function london_sounds_load in module mlend.processing:
- london_sounds_load(datadir_main=’../MLEnd/london_sounds’, train_test_split=’Benchmark_A’, verbose=1, encode_labels=True)
Read files of London Sounds Dataset and create training and testing sets.
- # Arguments
- datadir_main (str): local path where ‘MLEndLSD_audiofiles’ directory is stored
relative to ../MLEnd/london_sounds/).
- train_test_split (str): split type for training and testing
- ‘Benchmark_A’: Fixed Benchmark
Training (70%) and Testing (30%)
‘Random’ or ‘random’: random split woth 70-30
float (e.g. 0.8) (>0 and <1) random split with given fraction for training set. if train_test_split = 0.8, Training set will be 80% and Testing 20%
encode_labels: (bool), if to encode labels
- # Raises
- ValueError:
if train_test_split is not str [‘Benchmark_A’, ‘random’] or float (<1 and >0)”
- # Returns
TrainSet: A dictionary with keys {‘X_paths’, ‘Y’, ‘Y_encoded’} TestSet: A dictionary with keys {‘X_paths’, ‘Y’, ‘Y_encoded’}
‘X_paths’ is list of paths for audio files
‘Y’ is Nx3 np.array, column 0 for Area, 1 for Spot, and 2 for In_Out
- ‘Y_encoded’ is Nx3 np.array same as ‘Y’, column 0 for Area, 1 for Spot and 2 for In_Out
each column is encoded as 0, 1, 2 ..
MAPs : A dictionary of maps, if encode_labels is true, else an empty dictionary
mlend.yummy_small_load¶
Help on function yummy_small_load in module mlend.processing:
- yummy_small_load(datadir_main=’../MLEnd/yummy’, train_test_split=’Benchmark_A’, verbose=1, encode_labels=True)
Read files of Yummy Dataset and create training and testing sets.
- # Arguments
- datadir_main (str): local path where ‘MLEndHWD_audiofiles’ directory is stored
relative to ../MLEnd/hums_whistles/).
- train_test_split (str): split type for training and testing
- ‘Benchmark_A’: Speaker Independent Benchmark
Training (70%) and Testing (30%) do not have any common speaker
‘Random’ or ‘random’: random split woth 70-30
float (e.g. 0.8) (>0 and <1) random split with given fraction for training set. if train_test_split = 0.8, Training set will be 80% and Testing 20%
encode_labels: (bool), if to encode labels
- # Raises
- ValueError:
if train_test_split is not str [‘Benchmark_A’, ‘random’] or float (<1 and >0)”
- # Returns
TrainSet: A dictionary with keys {‘X_list’, ‘Y’, ‘Y_encoded’} TestSet: A dictionary with keys {‘X_list’, ‘Y’, ‘Y_encoded’}
‘X_paths’ is list of paths for audio files
‘Y’ is Nx1 np.array,
‘Y_encoded’ is Nx1 np.array same as ‘Y’, 0=rice 1=chips
MAPs : A dictionary of maps, if encode_labels is true, else an empty dictionary
mlend.yummy_load¶
Help on function yummy_load in module mlend.processing:
- yummy_load(datadir_main=’../MLEnd/yummy/’, train_test_split=’Benchmark_A’, verbose=1, attributes_as_labels=’all’, encode_labels=False)
Read files of Yummy Dataset and create training and testing sets.
- # Arguments
- datadir_main (str): local path where ‘MLEndYD_images’ directory is stored
relative to ../MLEnd/yummy/).
- train_test_split (str): split type for training and testing
- ‘Benchmark_A’: A predifined fixed split
Training (70%) and Testing (30%)
‘Random’ or ‘random’: random split woth 70-30
float (e.g. 0.8) (>0 and <1) random split with given fraction for training set. if train_test_split = 0.8, Training set will be 80% and Testing 20%
- attributes_as_labels: list of attribuetes as labels
attributes_as_labels = ‘all’ will return all the attribuetes as label
attributes_as_labels = [‘Diet’,’Healthiness_rating’] will return Y_train and Y_test as Nx2 columns diet and healthiness rating as labels
- encode_labels: (bool), if to encode labels
Only ‘Diet’, ‘Home_restaurent’, ‘Healthiness_rating’ and ‘Likeness’ will be encoded and return as numpy array
regardless of selection of attribuetes for labels
- # Raises
- ValueError:
if train_test_split is not str [‘Benchmark_A’, ‘random’] or float (<1 and >0)”
- # Returns
TrainSet: A dictionary with keys {‘X_list’, ‘Y’, ‘Y_encoded’} TestSet: A dictionary with keys {‘X_list’, ‘Y’, ‘Y_encoded’}
‘X_paths’ is list of paths for audio files
‘Y’ is NxC Pandas DataFrame,
‘Y_encoded’ is Nx4 np.array encoded labels for Diet, Home_or_restaurent, Healthiness and Likeness in that order.
MAPs : A dictionary of maps, if encode_labels is true, else an empty dictionary
mlend.happiness_load¶
Help on function happiness_load in module mlend.processing:
- happiness_load(datadir_main=’../MLEnd/happiness’, verbose=1, overwrite=False)
Read Happiness Dataset.
- # Arguments
- save_to: loacal path where you want to store the data
relative to ../MLEnd/happiness/).
- # Returns
Pandas Dataframe
Homepage
Fork me on Github
Examples Gellary