Fix typos discovered by codespell (microsoft#6)

deeprapper/README.md

@@ -85,7 +85,7 @@ To generate by our provided pretrained DeepRapper, first unzip the pretrained De
 │       └── config.json
 ```
 
-Finaly, run the following command to generate:
+Finally, run the following command to generate:
 ```bash
 bash generate_from_pretrain.sh
 ```

deeprapper/beam_search.py

@@ -258,7 +258,7 @@ def _control_rhymes(node, probs, tokenizer, beater, pinyin_dict, rhyme_words_lis
         rhyme_words = node.rhyme[0]
         for w in rhyme_words:
             probs = _rescale_rhymes(probs, w, tokenizer, beater, pinyin_dict, alpha)
-    else:  # detect the begining of new sentences
+    else:  # detect the beginning of new sentences
         last_token_id = node.wordid[0][0][-1]  # d2:batch size
         last_token = tokenizer.convert_ids_to_tokens([last_token_id])[0]
 

deeprapper/generate.py

@@ -92,9 +92,9 @@ def main():
     parser.add_argument('--reverse', action='store_true', help='whether to use reverse language model')
     parser.add_argument('--with_beat', action='store_true', help='whether to generate beats')
     parser.add_argument('--beat_mode', default=0, type=int, help='beat mode：0.no control；2.global；3.local', required=False)
-    parser.add_argument('--tempo', default=1, type=int, required=False, help='pace of beats:0-local controll; 1-slow; 2-medium; 3-fast')
+    parser.add_argument('--tempo', default=1, type=int, required=False, help='pace of beats:0-local control; 1-slow; 2-medium; 3-fast')
 
-    # beam seach param
+    # beam search param
     parser.add_argument('--beam_width', default=2, type=int, required=False, help='beam width')
     parser.add_argument('--beam_samples_num', default=5, type=int, required=False, help='beam searching samples')
     parser.add_argument('--beam_sample_select_sg', default='sample', type=str, required=False,

deeprapper/tokenizations/tokenization_bert.py

@@ -120,7 +120,7 @@ def __init__(self, vocab_file, do_lower_case=True, do_basic_tokenize=True, never
                 Only has an effect when do_basic_tokenize=True
             **tokenize_chinese_chars**: (`optional`) boolean (default True)
                 Whether to tokenize Chinese characters.
-                This should likely be desactivated for Japanese:
+                This should likely be deactivated for Japanese:
                 see: https://github.com/huggingface/pytorch-pretrained-BERT/issues/328
         """
         super(BertTokenizer, self).__init__(unk_token=unk_token, sep_token=sep_token,
@@ -215,7 +215,7 @@ def __init__(self, do_lower_case=True, never_split=None, tokenize_chinese_chars=
                 List of token not to split.
             **tokenize_chinese_chars**: (`optional`) boolean (default True)
                 Whether to tokenize Chinese characters.
-                This should likely be desactivated for Japanese:
+                This should likely be deactivated for Japanese:
                 see: https://github.com/huggingface/pytorch-pretrained-BERT/issues/328
         """
         if never_split is None:
@@ -401,7 +401,7 @@ def tokenize(self, text):
 
 def _is_whitespace(char):
     """Checks whether `chars` is a whitespace character."""
-    # \t, \n, and \r are technically contorl characters but we treat them
+    # \t, \n, and \r are technically control characters but we treat them
     # as whitespace since they are generally considered as such.
     if char == " " or char == "\t" or char == "\n" or char == "\r":
         return True

deeprapper/tokenizations/tokenization_bert_word_level.py

@@ -123,7 +123,7 @@ def __init__(self, vocab_file, do_lower_case=True, do_basic_tokenize=True, never
                 Only has an effect when do_basic_tokenize=True
             **tokenize_chinese_chars**: (`optional`) boolean (default True)
                 Whether to tokenize Chinese characters.
-                This should likely be desactivated for Japanese:
+                This should likely be deactivated for Japanese:
                 see: https://github.com/huggingface/pytorch-pretrained-BERT/issues/328
         """
         super(BertTokenizer, self).__init__(unk_token=unk_token, sep_token=sep_token,
@@ -218,7 +218,7 @@ def __init__(self, do_lower_case=True, never_split=None, tokenize_chinese_chars=
                 List of token not to split.
             **tokenize_chinese_chars**: (`optional`) boolean (default True)
                 Whether to tokenize Chinese characters.
-                This should likely be desactivated for Japanese:
+                This should likely be deactivated for Japanese:
                 see: https://github.com/huggingface/pytorch-pretrained-BERT/issues/328
         """
         if never_split is None:
@@ -416,7 +416,7 @@ def tokenize(self, text):
 
 def _is_whitespace(char):
     """Checks whether `chars` is a whitespace character."""
-    # \t, \n, and \r are technically contorl characters but we treat them
+    # \t, \n, and \r are technically control characters but we treat them
     # as whitespace since they are generally considered as such.
     if char == " " or char == "\t" or char == "\n" or char == "\r":
         return True

musicbert/binarize_genre.sh

@@ -6,7 +6,7 @@ PREFIX=$1
 [[ -d "${PREFIX}_data_bin" ]] && { echo "output directory ${PREFIX}_data_bin already exists" ; exit 1; }
 for i in {0..4}
 do
-	echo "Proccessing fold $i"
+	echo "Processing fold $i"
 	mkdir -p ${PREFIX}_data_bin/$i
 	fairseq-preprocess \
 	    --only-source \

pdaugment/README.md

@@ -132,7 +132,7 @@ metadata_dir            # metadata information of dataset, e.g., data/speech/pho
 output_duration_dir     # output directory of duration-augmented wav, e.g., data/duration
 output_pitch_dir        # output directory of pitch-augmented wav, e.g., data/pitch
 output_pdaugment_dir    # output directory of PDAugmented wav, e.g., data/pdaugment
-selected_dir            # select directory for training, validing and testing, e.g., ${output_pdaugment_dir}
+selected_dir            # select directory for training, validating and testing, e.g., ${output_pdaugment_dir}
 ```
 
 You can augment the speech dataset with multiple threads by setting:

pdaugment/midi_preprocess/steps/track_separate.py

@@ -13,7 +13,7 @@
 def remove_empty_track(midi_file):
     '''
     1. read pretty midi data
-    2. remove emtpy track,
+    2. remove empty track,
     also remove track with fewer than 10% notes of the track
     with most notes
     ********

pdaugment/midi_preprocess/utils/chord_recog.py

@@ -60,7 +60,7 @@
 
 
 def steps_per_bar_in_quantized_sequence(steps_per_quarter=4):
-    """Calculates steps per bar in a NoteSequence that has been quantized. assum time signature is 4/4
+    """Calculates steps per bar in a NoteSequence that has been quantized. assume time signature is 4/4
     Returns:
         Steps per bar as a floating point number.
         """

songmass/mass/song_sequence_generator.py

@@ -115,7 +115,7 @@ def _generate(
             raise Exception("expected src_tokens or source in net input")
 
         # bsz: total number of sentences in beam
-        # Note that src_tokens may have more than 2 dimenions (i.e. audio features)
+        # Note that src_tokens may have more than 2 dimensions (i.e. audio features)
         bsz, src_len = src_tokens.size()[:2]
         beam_size = self.beam_size
         # Get sentence numbers
@@ -176,7 +176,7 @@ def _generate(
         finalized = torch.jit.annotate(
             List[List[Dict[str, Tensor]]],
             [torch.jit.annotate(List[Dict[str, Tensor]], []) for i in range(bsz)],
-        )  # contains lists of dictionaries of infomation about the hypothesis being finalized at each step
+        )  # contains lists of dictionaries of information about the hypothesis being finalized at each step
 
         finished = [
             False for i in range(bsz)