pianistprogrammer commited on
Commit
de731d3
·
1 Parent(s): 747f6d0

Refactor Meter2800 dataset class by removing unused variables and simplifying example generation logic

Browse files
Files changed (1) hide show
  1. meter2800.py +11 -84
meter2800.py CHANGED
@@ -1,115 +1,42 @@
1
  import datasets
2
  import pandas as pd
3
- from pathlib import Path
4
-
5
- _CITATION = """\
6
- @misc{meter2800_dataset,
7
- author = {PianistProgrammer},
8
- title = {{Meter2800}: A Dataset for Music Time signature detection / Meter Classification},
9
- year = {2025},
10
- publisher = {Hugging Face},
11
- url = {https://huggingface.co/datasets/pianistprogrammer/Meter2800}
12
- }
13
- """
14
-
15
- _DESCRIPTION = """\
16
- Meter2800 is a dataset of 2,800 music audio samples for automatic meter classification.
17
- Each audio file is annotated with a primary meter class label and an alternative meter.
18
- It is split into training, validation, and test sets, each available in two class configurations:
19
- 2-class and 4-class. All audio is 16-bit WAV format.
20
- """
21
-
22
- _HOMEPAGE = "https://huggingface.co/datasets/pianistprogrammer/Meter2800"
23
- _LICENSE = "mit"
24
 
25
  class Meter2800(datasets.GeneratorBasedBuilder):
26
- """Meter2800 dataset for music meter classification."""
27
 
28
- VERSION = datasets.Version("1.0.0")
29
-
30
  BUILDER_CONFIGS = [
31
- datasets.BuilderConfig(
32
- name="4_classes",
33
- version=VERSION,
34
- description="4-class meter classification",
35
- ),
36
- datasets.BuilderConfig(
37
- name="2_classes",
38
- version=VERSION,
39
- description="2-class meter classification",
40
- ),
41
  ]
42
-
43
  DEFAULT_CONFIG_NAME = "4_classes"
44
 
45
  def _info(self):
46
- # We'll determine the labels dynamically from the CSV files
47
- # For now, use a generic ClassLabel that will be updated later
48
  return datasets.DatasetInfo(
49
- description=_DESCRIPTION,
50
  features=datasets.Features({
51
  "filename": datasets.Value("string"),
52
  "audio": datasets.Audio(sampling_rate=None),
53
- "label": datasets.Value("string"), # We'll convert to ClassLabel later
54
  "meter": datasets.Value("string"),
55
  "alt_meter": datasets.Value("string"),
56
  }),
57
- supervised_keys=("audio", "label"),
58
- homepage=_HOMEPAGE,
59
- license=_LICENSE,
60
- citation=_CITATION,
61
  )
62
 
63
  def _split_generators(self, dl_manager):
64
- """Returns SplitGenerators."""
65
-
66
- # The files should be in the root of the repo
67
- config_suffix = self.config.name
68
-
69
  return [
70
  datasets.SplitGenerator(
71
  name=datasets.Split.TRAIN,
72
- gen_kwargs={
73
- "csv_file": f"data_train_{config_suffix}.csv",
74
- "split": "train",
75
- },
76
- ),
77
- datasets.SplitGenerator(
78
- name=datasets.Split.VALIDATION,
79
- gen_kwargs={
80
- "csv_file": f"data_val_{config_suffix}.csv",
81
- "split": "validation",
82
- },
83
- ),
84
- datasets.SplitGenerator(
85
- name=datasets.Split.TEST,
86
- gen_kwargs={
87
- "csv_file": f"data_test_{config_suffix}.csv",
88
- "split": "test",
89
- },
90
  ),
91
  ]
92
 
93
- def _generate_examples(self, csv_file, split):
94
- """Yields examples."""
95
-
96
- # Read CSV directly - the file should be available in the repo
97
- df = pd.read_csv(csv_file)
98
-
99
- # Read CSV
100
  df = pd.read_csv(csv_file)
101
- df = df.dropna(subset=["filename", "label"]).reset_index(drop=True)
102
-
103
  for idx, row in df.iterrows():
104
- # The audio files should be in subdirectories
105
- audio_file = row["filename"]
106
- if not audio_file.startswith("/"):
107
- audio_file = "/" + audio_file
108
-
109
  yield idx, {
110
  "filename": row["filename"],
111
- "audio": audio_file.lstrip("/"), # Remove leading slash for HF
112
- "label": str(row["label"]),
113
- "meter": str(row.get("meter", "")),
114
- "alt_meter": str(row.get("alt_meter", row.get("meter", ""))),
115
  }
 
1
  import datasets
2
  import pandas as pd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
 
4
  class Meter2800(datasets.GeneratorBasedBuilder):
5
+ """Minimal test version."""
6
 
 
 
7
  BUILDER_CONFIGS = [
8
+ datasets.BuilderConfig(name="4_classes", description="4 class version"),
9
+ datasets.BuilderConfig(name="2_classes", description="2 class version"),
 
 
 
 
 
 
 
 
10
  ]
11
+
12
  DEFAULT_CONFIG_NAME = "4_classes"
13
 
14
  def _info(self):
 
 
15
  return datasets.DatasetInfo(
 
16
  features=datasets.Features({
17
  "filename": datasets.Value("string"),
18
  "audio": datasets.Audio(sampling_rate=None),
19
+ "label": datasets.Value("string"),
20
  "meter": datasets.Value("string"),
21
  "alt_meter": datasets.Value("string"),
22
  }),
 
 
 
 
23
  )
24
 
25
  def _split_generators(self, dl_manager):
 
 
 
 
 
26
  return [
27
  datasets.SplitGenerator(
28
  name=datasets.Split.TRAIN,
29
+ gen_kwargs={"csv_file": f"data_train_{self.config.name}.csv"},
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30
  ),
31
  ]
32
 
33
+ def _generate_examples(self, csv_file):
 
 
 
 
 
 
34
  df = pd.read_csv(csv_file)
 
 
35
  for idx, row in df.iterrows():
 
 
 
 
 
36
  yield idx, {
37
  "filename": row["filename"],
38
+ "audio": row["filename"].lstrip("/"),
39
+ "label": row["label"],
40
+ "meter": str(row["meter"]),
41
+ "alt_meter": str(row["alt_meter"]),
42
  }