Update magnatagatune.py

magnatagatune.py

@@ -1,6 +1,6 @@
 # coding=utf-8
 
-"""Medley-Solos-DB dataset."""
+"""MagnaTagATune dataset."""
 
 
 import os
@@ -18,13 +18,14 @@ import urllib.request
 from pathlib import Path
 from copy import deepcopy
 from tqdm.auto import tqdm
+from rich import print
 from rich.logging import RichHandler
 
 logger = logging.getLogger(__name__)
 logger.addHandler(RichHandler())
 logger.setLevel(logging.INFO)
 
-SAMPLE_RATE = 44_100
+SAMPLE_RATE = 16_000
 
 # Cache location
 VERSION = "0.0.1"
@@ -36,38 +37,41 @@ DEFAULT_HF_DATASETS_CACHE = os.path.join(HF_CACHE_HOME, "datasets")
 HF_DATASETS_CACHE = Path(os.getenv("HF_DATASETS_CACHE", DEFAULT_HF_DATASETS_CACHE))
 
 CLASSES = [
-    'clarinet', 'distorted electric guitar', 'female singer', 'flute', 'piano', 'tenor saxophone', 'trumpet', 'violin'
+    "guitar", "classical", "slow", "techno", "strings", "drums", "electronic", "rock", "fast", "piano", 
+    "ambient", "beat", "violin", "vocal", "synth", "female", "indian", "opera", "male", "singing", "vocals", 
+    "no vocals", "harpsichord", "loud", "quiet", "flute", "woman", "male vocal", "no vocal", "pop", "soft", 
+    "sitar", "solo", "man", "classic", "choir", "voice", "new age", "dance", "male voice", "female vocal", 
+    "beats", "harp", "cello", "no voice", "weird", "country", "metal", "female voice", "choral"
 ]
-CLASS2INDEX = {cls:idx for idx, cls in enumerate(CLASSES)}
-INDEX2CLASS = {idx:cls for idx, cls in enumerate(CLASSES)}
+CLASSES = sorted(CLASSES)
 
 
-class MedleySolosDBConfig(datasets.BuilderConfig):
-    """BuilderConfig for Medley-Solos-DB."""
+class MagnaTagATuneConfig(datasets.BuilderConfig):
+    """BuilderConfig for MagnaTagATune."""
     
     def __init__(self, features, **kwargs):
-        super(MedleySolosDBConfig, self).__init__(version=datasets.Version(VERSION, ""), **kwargs)
+        super(MagnaTagATuneConfig, self).__init__(version=datasets.Version(VERSION, ""), **kwargs)
         self.features = features
 
 
-class MedleySolosDB(datasets.GeneratorBasedBuilder):
+class MagnaTagATune(datasets.GeneratorBasedBuilder):
 
     BUILDER_CONFIGS = [
-        MedleySolosDBConfig(
+        MagnaTagATuneConfig(
             features=datasets.Features(
                 {
                     "file": datasets.Value("string"),
                     "audio": datasets.Audio(sampling_rate=SAMPLE_RATE),
-                    "instrument": datasets.Value("string"), 
-                    "label": datasets.features.ClassLabel(names=CLASSES), 
+                    "sound": datasets.Sequence(datasets.Value("string")), 
+                    "label": datasets.Sequence(datasets.features.ClassLabel(names=CLASSES)), 
                 }
             ),
-            name="v1.2", 
+            name="top50", 
             description="",
         ), 
     ]
 
-    DEFAULT_CONFIG_NAME = "v1.2"
+    DEFAULT_CONFIG_NAME = "top50"
 
     def _info(self):
         return datasets.DatasetInfo(
@@ -81,10 +85,10 @@ class MedleySolosDB(datasets.GeneratorBasedBuilder):
 
     def _split_generators(self, dl_manager):
         """Returns SplitGenerators."""
-        zip_file_url = "https://zenodo.org/records/3464194/files/Medley-solos-DB.tar.gz"
+        zip_file_url = "https://huggingface.co/datasets/confit/magnatagatune/resolve/main/mp3.zip"
         _filename = zip_file_url.split('/')[-1]
         _save_path = os.path.join(
-            HF_DATASETS_CACHE, 'confit___medley-solos-db/v1.2', VERSION, _filename
+            HF_DATASETS_CACHE, 'confit___magnatagatune/top50', VERSION, _filename
         )
         download_file(zip_file_url, _save_path)
         logger.info(f"`{_filename}` is downloaded to {_save_path}")
@@ -104,46 +108,65 @@ class MedleySolosDB(datasets.GeneratorBasedBuilder):
         ]
 
     def _generate_examples(self, archive_path, split=None):
-        metadata_df = pd.read_csv("https://zenodo.org/records/3464194/files/Medley-solos-DB_metadata.csv")
-        train_df = metadata_df[metadata_df["subset"] == "training"].reset_index(drop=True)
-        validation_df = metadata_df[metadata_df["subset"] == "validation"].reset_index(drop=True)
-        test_df = metadata_df[metadata_df["subset"] == "test"].reset_index(drop=True)
+        df = pd.read_csv(
+            'https://huggingface.co/datasets/confit/magnatagatune/resolve/main/annotations_final.csv', sep="\t"
+        )
+        # Filter only the songs that have at least one of the top 50 tags
+        df = df[df[CLASSES].sum(axis=1) > 0].reset_index(drop=True)
+        df = df[CLASSES + ["mp3_path", "clip_id"]]
+        train_ids_df = pd.read_csv(
+            "https://huggingface.co/datasets/confit/magnatagatune/resolve/main/train_gt_mtt.tsv", sep="\t", header=None
+        )
+        train_ids = train_ids_df[0].tolist()
+        train_df = df[df["clip_id"].isin(train_ids)].reset_index(drop=True)
+
+        validation_ids_df = pd.read_csv(
+            "https://huggingface.co/datasets/confit/magnatagatune/resolve/main/val_gt_mtt.tsv", sep="\t", header=None
+        )
+        validation_ids = validation_ids_df[0].tolist()
+        validation_df = df[df["clip_id"].isin(validation_ids)].reset_index(drop=True)
 
-        extensions = ['.wav']
+        test_ids_df = pd.read_csv(
+            "https://huggingface.co/datasets/confit/magnatagatune/resolve/main/test_gt_mtt.tsv", sep="\t", header=None
+        )
+        test_ids = test_ids_df[0].tolist()
+        test_df = df[df["clip_id"].isin(test_ids)].reset_index(drop=True)
+
+        extensions = ['.mp3']
         _, _walker = fast_scandir(archive_path, extensions, recursive=True)
 
+        # Extract the list of column names where the value is 1 for each row
+        result = df.apply(lambda row: [col for col in df.columns if row[col] == 1], axis=1).tolist()
+
         if split == 'train':
             fileid2class = {}
             for idx, row in train_df.iterrows():
-                fileid = row['uuid4']
-                class_ = row['instrument']
+                fileid = os.path.join(archive_path, str(row['mp3_path']))
+                class_ = result[idx]
                 fileid2class[fileid] = class_
         elif split == 'validation':
             fileid2class = {}
             for idx, row in validation_df.iterrows():
-                fileid = row['uuid4']
-                class_ = row['instrument']
+                fileid = os.path.join(archive_path, str(row['mp3_path']))
+                class_ = result[idx]
                 fileid2class[fileid] = class_
         elif split == 'test':
             fileid2class = {}
             for idx, row in test_df.iterrows():
-                fileid = row['uuid4']
-                class_ = row['instrument']
+                fileid = os.path.join(archive_path, str(row['mp3_path']))
+                class_ = result[idx]
                 fileid2class[fileid] = class_
         
-        _walker = [fileid for fileid in _walker if not Path(fileid).name.startswith('._Medley')]
         for guid, audio_path in enumerate(_walker):
-            fileid = Path(audio_path).stem
-            fileid = fileid.split('_')[-1]
-            if fileid not in fileid2class:
+            if audio_path not in fileid2class:
                 continue
-            instrument = fileid2class.get(fileid)
+            tags = fileid2class.get(audio_path)
             yield guid, {
                 "id": str(guid), 
                 "file": audio_path, 
                 "audio": audio_path, 
-                "instrument": instrument, 
-                "label": instrument, 
+                "sound": tags, 
+                "label": tags, 
             }