Clean up (#175)

Author: sjkoelle

settings.ini

@@ -27,7 +27,7 @@ status = 2
 # Optional. Same format as setuptools requirements.  Torch version seems to effect random number generator (not 100% certain).
 # TODO (Sam): our goal is to rely on as few 3rd party packages as possible.  We should try to remove as many of these as possible and integrate torch code directly.
 # NOTE (Sam): is it possible to specify no-deps here?
-requirements = Cython pytest phonemizer inflect librosa>=0.8.0 matplotlib nltk>=3.6.5 numpy>=1.23.5 csvw clldutils pandas pydub scipy scikit-learn soundfile tensorboardX torch>=1.13.0 torchaudio>=0.9.0 unidecode seaborn mdutils wordcloud wordfreq Pillow einops g2p_en@git+https://github.com/uberduck-ai/g2p emoji text-unidecode gdown pre-commit lmdb  ray[default] praat-parselmouth>=0.4.3
+requirements = Cython pytest phonemizer inflect librosa>=0.8.0 matplotlib nltk>=3.6.5 numpy>=1.23.5 pandas pydub scipy scikit-learn tensorboardX torch>=1.13.0 torchaudio>=0.9.0 unidecode seaborn wordfreq einops g2p_en@git+https://github.com/uberduck-ai/g2p emoji text-unidecode pre-commit lmdb ray[default] praat-parselmouth>=0.4.3
 
 # Optional. Same format as setuptools console_scripts
 # console_scripts =

uberduck_ml_dev/data/statistics.py

@@ -13,12 +13,8 @@
 import os
 
 import librosa
-import matplotlib.pyplot as plt
 import numpy as np
-from PIL import Image, ImageOps
 from pydub.utils import mediainfo_json
-import seaborn as sns
-from wordcloud import WordCloud, STOPWORDS
 from wordfreq import word_frequency
 
 from ..text.utils import text_to_sequence
@@ -36,33 +32,6 @@ def word_frequencies(text: str, language: str = "en") -> List[float]:
     return freqs
 
 
-def create_wordcloud(text: str, output_file: str):
-    """
-    Creates and saves a wordcloud in the shape of a duck to the designated output file.
-    """
-    dir_path = os.path.dirname(os.path.realpath(__file__))
-
-    mask = np.array(
-        ImageOps.invert(
-            Image.open(os.path.join(dir_path, "../assets/duck.png")).convert("RGB")
-        )
-    )
-    wc = WordCloud(
-        background_color="white",
-        max_words=3000,
-        mask=mask,
-        contour_width=7,
-        contour_color="steelblue",
-    )
-    frequency_dict = dict(Counter(text.lower().split()).most_common())
-    for k in STOPWORDS:
-        if k in frequency_dict.keys():
-            frequency_dict.pop(k)
-
-    wc.generate_from_frequencies(frequency_dict)
-    wc.to_file(output_file)
-
-
 def count_frequency(arr: List[Any]) -> Dict[Any, int]:
     """
     Calculates the frequency that a value appears in a list