New version of CompMusic Turkish Makam Dataset

mirdata/datasets/compmusic_otmm_makam.py

@@ -3,6 +3,11 @@
 .. admonition:: Dataset Info
     :class: dropdown
 
+    **NOTE**: From mirdata v0.3.8 on, the only version available of this dataset is dlfm2016-fix1, which is
+    basically the same as dlfm2016, but with a few fixes in some annotations. The original dlfm2016 version
+    is still available in mirdata versions <=0.3.7. Note that from dlfm2016 to dlfm2016-fix1, no new recordings
+    or annotation were added, only a few annotation files were fixed.
+
     This dataset is designed to test makam recognition methodologies on Ottoman-Turkish makam music.
     It is composed of 50 recording from each of the 20 most common makams in CompMusic Project's Dunya Ottoman-Turkish
     Makam Music collection. Currently the dataset is the largest makam recognition dataset.
@@ -58,16 +63,18 @@
 """
 
 INDEXES = {
-    "default": "dlfm2016",
-    "test": "dlfm2016",
-    "dlfm2016": core.Index(filename="compmusic_otmm_makam_index_dlfm2016.json"),
+    "default": "dlfm2016-fix1",
+    "test": "dlfm2016-fix1",
+    "dlfm2016-fix1": core.Index(
+        filename="compmusic_otmm_makam_index_dlfm2016-fix1.json"
+    ),
 }
 
 REMOTES = {
     "all": download_utils.RemoteFileMetadata(
-        filename="otmm_makam_recognition_dataset-dlfm2016.zip",
-        url="https://zenodo.org/record/58413/files/otmm_makam_recognition_dataset-dlfm2016.zip?download=1",
-        checksum="c2b9c8bdcbdcf15745b245adfc793145",
+        filename="otmm_makam_recognition_dataset-dlfm2016-fix1.zip",
+        url="https://zenodo.org/record/4883680/files/MTG/otmm_makam_recognition_dataset-dlfm2016-fix1.zip?download=1",
+        checksum="83724c889d36f684cff3f15f20ce0d34",
     )
 }
 
@@ -176,7 +183,10 @@ def load_mb_tags(fhandle: TextIO) -> dict:
         Dict: metadata of the track
 
     """
-    return json.load(fhandle)
+    mb_tags = json.load(fhandle)
+    if "duration" not in mb_tags.keys():
+        mb_tags["duration"] = 0.0  # Few tracks have no duration information
+    return mb_tags
 
 
 @core.docstring_inherit(core.Dataset)
@@ -201,7 +211,7 @@ def __init__(self, data_home=None, version="default"):
     def _metadata(self):
         metadata_path = os.path.join(
             os.path.normpath(self.data_home),
-            "MTG-otmm_makam_recognition_dataset-f14c0d0",
+            "MTG-otmm_makam_recognition_dataset-55ce75a",
             "annotations.json",
         )
 

scripts/make_compmusic_otmm_makam_index.py

@@ -5,12 +5,12 @@
 from mirdata.validate import md5
 
 
-OTMM_MAKAM_INDEX_PATH = '../mirdata/datasets/indexes/otmm_makam_index.json'
+OTMM_MAKAM_INDEX_PATH = '../mirdata/datasets/indexes/compmusic_otmm_makam_index_dlfm2016-fix1.json'
 
 
 def make_otmm_makam_index(dataset_data_path):
 
-    otmm_index = {'version': 'dlfm2016', 'tracks': {}, 'metadata': []}
+    otmm_index = {'version': 'dlfm2016-fix1', 'tracks': {}, 'metadata': []}
 
     for makam in os.listdir(os.path.join(dataset_data_path, 'data')):
         if '.' not in makam:
@@ -23,7 +23,7 @@ def make_otmm_makam_index(dataset_data_path):
                     otmm_index['tracks'][index] = {
                         "metadata": [
                             os.path.join(
-                                'MTG-otmm_makam_recognition_dataset-f14c0d0',
+                                'MTG-otmm_makam_recognition_dataset-55ce75a',
                                 'data',
                                 makam,
                                 track,
@@ -32,7 +32,7 @@ def make_otmm_makam_index(dataset_data_path):
                         ],
                         "pitch": [
                             os.path.join(
-                                'MTG-otmm_makam_recognition_dataset-f14c0d0',
+                                'MTG-otmm_makam_recognition_dataset-55ce75a',
                                 'data',
                                 makam,
                                 pitch_path,
@@ -44,10 +44,12 @@ def make_otmm_makam_index(dataset_data_path):
                             ),
                         ],
                     }
-    otmm_index['metadata'] = [
-        os.path.join('MTG-otmm_makam_recognition_dataset-f14c0d0', 'annotations.json'),
-        md5(os.path.join(dataset_data_path, 'annotations.json')),
-    ]
+    otmm_index['metadata'] = {
+        "annotation_metadata": [
+            os.path.join('MTG-otmm_makam_recognition_dataset-55ce75a', 'annotations.json'),
+            md5(os.path.join(dataset_data_path, 'annotations.json')),
+        ]
+    }
 
     with open(OTMM_MAKAM_INDEX_PATH, 'w') as fhandle:
         json.dump(otmm_index, fhandle, indent=2)

tests/datasets/test_otmm_makam.py

@@ -16,11 +16,11 @@ def test_track():
         "track_id": "cafcdeaf-e966-4ff0-84fb-f660d2b68365",
         "pitch_path": os.path.join(
             os.path.normpath("tests/resources/mir_datasets/compmusic_otmm_makam/"),
-            "MTG-otmm_makam_recognition_dataset-f14c0d0/data/Kurdilihicazkar/cafcdeaf-e966-4ff0-84fb-f660d2b68365.pitch",
+            "MTG-otmm_makam_recognition_dataset-55ce75a/data/Kurdilihicazkar/cafcdeaf-e966-4ff0-84fb-f660d2b68365.pitch",
         ),
         "mb_tags_path": os.path.join(
             os.path.normpath("tests/resources/mir_datasets/compmusic_otmm_makam/"),
-            "MTG-otmm_makam_recognition_dataset-f14c0d0/data/Kurdilihicazkar/cafcdeaf-e966-4ff0-84fb-f660d2b68365.json",
+            "MTG-otmm_makam_recognition_dataset-55ce75a/data/Kurdilihicazkar/cafcdeaf-e966-4ff0-84fb-f660d2b68365.json",
         ),
         "form": "sarki",
         "instrumentation": "Solo vocal with accompaniment",
@@ -707,6 +707,14 @@ def test_load_mb_tags():
         }
     ]
 
+    # test empty duration
+    track_id = "2a3e1a6d-ab82-4a46-ae8b-0fb057b53de0"
+    dataset = compmusic_otmm_makam.Dataset(data_home)
+    track = dataset.track(track_id)
+    mb_tags_path = track.mb_tags_path
+    mb_tags = compmusic_otmm_makam.load_mb_tags(mb_tags_path)
+    assert mb_tags["duration"] == 0.0
+
 
 def test_special_turkish_characters():
     data_home = "tests/resources/mir_datasets/compmusic_otmm_makam"

tests/resources/mir_datasets/compmusic_otmm_makam/MTG-otmm_makam_recognition_dataset-55ce75a/data/Segah/2a3e1a6d-ab82-4a46-ae8b-0fb057b53de0.json

@@ -0,0 +1,14 @@
+{
+    "usul": [
+        {
+            "attribute_key": "aksak", 
+            "mb_attribute": "Aksak", 
+            "source": "http://musicbrainz.org/work/753ff394-dec1-422b-991f-227d8f848532"
+        }, 
+        {
+            "attribute_key": "aksak", 
+            "mb_tag": "aksak", 
+            "source": "http://musicbrainz.org/recording/cafcdeaf-e966-4ff0-84fb-f660d2b68365"
+        }
+    ]
+}