Update esc-datasets.py

esc-datasets.py

@@ -32,8 +32,6 @@ from pathlib import Path
 from huggingface_hub import HfApi, HfFolder
 import datasets
 
-from .cv_release_stats import STATS as _COMMON_VOICE_STATS
-
 
 _DESCRIPTIONS = {
     "ami": """
@@ -391,7 +389,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
 
                 transcriptions[audio_filename] = {
                     "id": _id,
-                    "text": text,
+                    "text": text if split != "eval" else "",
                 }
 
         features = ["id", "text"]
@@ -438,6 +436,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "local_extracted_archive_paths": local_extracted_archive_paths[subconfig],
                     "archives": [dl_manager.iter_archive(path) for path in archive_paths[subconfig]],
                     "meta_path": meta_path[subconfig],
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -446,6 +445,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "local_extracted_archive_paths": local_extracted_archive_paths["dev"],
                     "archives": [dl_manager.iter_archive(path) for path in archive_paths["dev"]],
                     "meta_path": meta_path["dev"],
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -454,11 +454,12 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "local_extracted_archive_paths": local_extracted_archive_paths["test"],
                     "archives": [dl_manager.iter_archive(path) for path in archive_paths["test"]],
                     "meta_path": meta_path["test"],
+                    "is_test": True,
                 },
             ),
         ]
 
-    def _spgispeech_generate_examples(self, local_extracted_archive_paths, archives, meta_path):
+    def _spgispeech_generate_examples(self, local_extracted_archive_paths, archives, meta_path, is_test):
         # define the expected metadata dict keys,
         # some files have metadata with erroneous entries that we have to filter out
         dict_keys = {"id": "wav_filename", "text": "transcript"}
@@ -482,6 +483,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 # get the .wav filename by removing the directory path from the audio filename
                 wav_filename = "/".join(audio_filename.split("/")[-2:])
                 example = dict(metadata[wav_filename])
+                if is_test: example["text"] = ""
                 example["audio"] = {"path": path, "bytes": audio_file.read()}
                 example["dataset"] = "spgispeech"
                 yield audio_filename, example
@@ -502,7 +504,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
             split: _VOXPOPULI_METADATA_PATH.format(split=split) for split in splits
         }
 
-        dl_manager.download_config.num_proc = len(audio_urls["train"])
+        dl_manager.download_config.num_proc = len(audio_urls["train"]) // 4
         meta_paths = dl_manager.download_and_extract(meta_urls)
         audio_paths = dl_manager.download(audio_urls)
 
@@ -519,6 +521,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_paths["train"]],
                     "local_extracted_archives_paths": local_extracted_audio_paths["train"],
                     "meta_path": meta_paths["train"],
+                    "is_test": False,
                 }
             ),
             datasets.SplitGenerator(
@@ -527,6 +530,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_paths["dev"]],
                     "local_extracted_archives_paths": local_extracted_audio_paths["dev"],
                     "meta_path": meta_paths["dev"],
+                    "is_test": False,
                 }
             ),
             datasets.SplitGenerator(
@@ -535,11 +539,12 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_paths["test"]],
                     "local_extracted_archives_paths": local_extracted_audio_paths["test"],
                     "meta_path": meta_paths["test"],
+                    "is_test": True,
                 }
             ),
         ]
 
-    def _voxpopuli_generate_examples(self, audio_archives, local_extracted_archives_paths, meta_path):
+    def _voxpopuli_generate_examples(self, audio_archives, local_extracted_archives_paths, meta_path, is_test):
         assert len(audio_archives) == len(local_extracted_archives_paths)
 
         logging.info("Reading voxpopuli metadata.")
@@ -553,7 +558,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
 
                 yield audio_id, {
                     "id": audio_id,
-                    "text": metadata[audio_id]["normalized_text"].lower(),
+                    "text": metadata[audio_id]["normalized_text"].lower() if not is_test else "",
                     "audio": {"path": path, "bytes": audio_file.read()},
                     "dataset": "voxpopuli",
                 }
@@ -572,6 +577,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 gen_kwargs={
                     "local_extracted_archives": [local_extracted_archives.get(train_name) for train_name in train_splits],
                     "archives": [dl_manager.iter_archive(archive_paths[train_name]) for train_name in train_splits],
+                    "is_test": False,
                 },
             )
         ]
@@ -581,6 +587,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 gen_kwargs={
                     "local_extracted_archives": [local_extracted_archives.get("dev.clean")],
                     "archives": [dl_manager.iter_archive(archive_paths["dev.clean"])],
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -588,6 +595,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 gen_kwargs={
                     "local_extracted_archives": [local_extracted_archives.get("dev.other")],
                     "archives": [dl_manager.iter_archive(archive_paths["dev.other"])],
+                    "is_test": False,
                 },
             ),
         ]
@@ -597,6 +605,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 gen_kwargs={
                     "local_extracted_archives": [local_extracted_archives.get("test.clean")],
                     "archives": [dl_manager.iter_archive(archive_paths["test.clean"])],
+                    "is_test": True,
                 },
             ),
             datasets.SplitGenerator(
@@ -604,12 +613,13 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 gen_kwargs={
                     "local_extracted_archives": [local_extracted_archives.get("test.other")],
                     "archives": [dl_manager.iter_archive(archive_paths["test.other"])],
+                    "is_test": True,
                 },
             ),
         ]
         return train_split + dev_splits + test_splits
 
-    def _librispeech_generate_examples(self, archives, local_extracted_archives):
+    def _librispeech_generate_examples(self, archives, local_extracted_archives, is_test):
         key = 0
         audio_data = {}
         transcripts = []
@@ -637,7 +647,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                                 {
                                     "id": id_,
                                     "file": audio_file,
-                                    "text": transcript,
+                                    "text": transcript if not is_test else "",
                                 }
                             )
                 if audio_data and len(audio_data) == len(transcripts):
@@ -674,12 +684,11 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 "Please set use_auth_token=True or use_auth_token='<TOKEN>' to download this dataset"
             )
 
-        bundle_url_template = _COMMON_VOICE_STATS["bundleURLTemplate"]
-        bundle_version = bundle_url_template.split("/")[0]
+        bundle_version = _COMMON_VOICE_BUNDLE_URL_TEMPLATE.split("/")[0]
         dl_manager.download_config.ignore_url_params = True
 
         self._common_voice_log_download("en", bundle_version, hf_auth_token)
-        archive_path = dl_manager.download(self._common_voice_get_bundle_url("en", bundle_url_template))
+        archive_path = dl_manager.download(self._common_voice_get_bundle_url("en", _COMMON_VOICE_BUNDLE_URL_TEMPLATE))
         local_extracted_archive = dl_manager.extract(archive_path) if not dl_manager.is_streaming else None
 
         path_to_data = "/".join([bundle_version, "en"])
@@ -693,15 +702,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "archive_iterator": dl_manager.iter_archive(archive_path),
                     "metadata_filepath": "/".join([path_to_data, "train.tsv"]) if path_to_data else "train.tsv",
                     "path_to_clips": path_to_clips,
-                },
-            ),
-            datasets.SplitGenerator(
-                name=datasets.Split.TEST,
-                gen_kwargs={
-                    "local_extracted_archive": local_extracted_archive,
-                    "archive_iterator": dl_manager.iter_archive(archive_path),
-                    "metadata_filepath": "/".join([path_to_data, "test.tsv"]) if path_to_data else "test.tsv",
-                    "path_to_clips": path_to_clips,
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -711,26 +712,17 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "archive_iterator": dl_manager.iter_archive(archive_path),
                     "metadata_filepath": "/".join([path_to_data, "dev.tsv"]) if path_to_data else "dev.tsv",
                     "path_to_clips": path_to_clips,
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
-                name="other",
-                gen_kwargs={
-                    "local_extracted_archive": local_extracted_archive,
-                    "archive_iterator": dl_manager.iter_archive(archive_path),
-                    "metadata_filepath": "/".join([path_to_data, "other.tsv"]) if path_to_data else "other.tsv",
-                    "path_to_clips": path_to_clips,
-                },
-            ),
-            datasets.SplitGenerator(
-                name="invalidated",
+                name=datasets.Split.TEST,
                 gen_kwargs={
                     "local_extracted_archive": local_extracted_archive,
                     "archive_iterator": dl_manager.iter_archive(archive_path),
-                    "metadata_filepath": "/".join([path_to_data, "invalidated.tsv"])
-                    if path_to_data
-                    else "invalidated.tsv",
+                    "metadata_filepath": "/".join([path_to_data, "test.tsv"]) if path_to_data else "test.tsv",
                     "path_to_clips": path_to_clips,
+                    "is_test": True,
                 },
             ),
         ]
@@ -741,6 +733,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
         archive_iterator,
         metadata_filepath,
         path_to_clips,
+        is_test,
     ):
         """Yields examples."""
         data_fields = list(self._info().features.keys())
@@ -783,7 +776,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                         text = text[1:-1]
                     # replace double quotation marks with single
                     text = text.replace('""', '"')
-                    result["text"] = text
+                    result["text"] = text if not is_test else ""
 
                     yield path, result
 
@@ -850,7 +843,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                             key = "-".join([speaker, start, end, label])
                             example = {
                                 "audio": {"path": audio_file, "array": samples, "sampling_rate": sampling_rate},
-                                "text": transcript,
+                                "text": transcript if split_path != "test" else "",
                                 "id": key,
                                 "dataset": "tedlium",
                             }
@@ -920,7 +913,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                                 key = transcript["id"]
                                 yield key, {
                                     "audio": audio,
-                                    "text": transcript["text"],
+                                    "text": transcript["text"] if split_path != "test" else "",
                                     "dataset": "tedlium",
                                     "id": transcript["id"],
                                 }
@@ -969,7 +962,8 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                         dl_manager.iter_archive(archive_path) for archive_path in audio_archives_paths["train"]
                     ],
                     "local_audio_archives_paths": local_audio_archives_paths.get("train"),
-                    "meta_paths": meta_paths["train"]
+                    "meta_paths": meta_paths["train"],
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -979,7 +973,8 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                         dl_manager.iter_archive(archive_path) for archive_path in audio_archives_paths["dev"]
                     ],
                     "local_audio_archives_paths": local_audio_archives_paths.get("dev"),
-                    "meta_paths": meta_paths["dev"]
+                    "meta_paths": meta_paths["dev"],
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -989,12 +984,13 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                         dl_manager.iter_archive(archive_path) for archive_path in audio_archives_paths["test"]
                     ],
                     "local_audio_archives_paths": local_audio_archives_paths.get("test"),
-                    "meta_paths": meta_paths["test"]
+                    "meta_paths": meta_paths["test"],
+                    "is_test": True,
                 },
             ),
         ]
 
-    def _gigaspeech_generate_examples(self, audio_archives_iterators, local_audio_archives_paths, meta_paths):
+    def _gigaspeech_generate_examples(self, audio_archives_iterators, local_audio_archives_paths, meta_paths, is_test):
         assert len(audio_archives_iterators) == len(meta_paths)
         if local_audio_archives_paths:
             assert len(audio_archives_iterators) == len(local_audio_archives_paths)
@@ -1031,7 +1027,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                 if len(text) == 0:
                     continue
 
-                audio_meta["text"] = text
+                audio_meta["text"] = text if not is_test else ""
 
                 path = os.path.join(local_audio_archives_paths[i], audio_path_in_archive) if local_audio_archives_paths \
                     else audio_path_in_archive
@@ -1074,6 +1070,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "local_extracted_archive_paths": local_extracted_archive_paths["train"],
                     "archives": [dl_manager.iter_archive(path) for path in archive_paths["train"]],
                     "metadata": metadata,
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -1082,6 +1079,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "local_extracted_archive_paths": local_extracted_archive_paths["dev"],
                     "archives": [dl_manager.iter_archive(path) for path in archive_paths["dev"]],
                     "metadata": metadata,
+                    "is_test": False,
                 },
             ),
             datasets.SplitGenerator(
@@ -1090,11 +1088,12 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
                     "local_extracted_archive_paths": local_extracted_archive_paths["test"],
                     "archives": [dl_manager.iter_archive(path) for path in archive_paths["test"]],
                     "metadata": metadata,
+                    "is_test": True,
                 },
             ),
         ]
 
-    def _earnings_generate_examples(self, local_extracted_archive_paths, archives, metadata):
+    def _earnings_generate_examples(self, local_extracted_archive_paths, archives, metadata, is_test):
         for local_extracted_archive_path, archive in zip(local_extracted_archive_paths, archives):
             # Here we iterate over all the files within the TAR archive:
             for audio_filename, audio_file in archive:
@@ -1124,7 +1123,7 @@ class ESCDatasets(datasets.GeneratorBasedBuilder):
 
                 yield audio_filename, {
                     "id": audio_filename,
-                    "text": text,
+                    "text": text if not is_test else "",
                     "dataset": "earnings22",
                     "audio": {"path": path, "bytes": audio_file.read()}
                 }
@@ -1366,7 +1365,7 @@ _AMI_ANNOTATIONS_ARCHIVE_URL = _AMI_BASE_DATA_URL + "annotations/{split}/text"
 
 _SPGISPEECH_BASE_URL = "https://huggingface.co/datasets/kensho/spgispeech/resolve/main/data/"
 
-_SPGISPEECH_AUDIO_BASE_URL = _SPGISPEECH_BASE_URL + "/audio"
+_SPGISPEECH_AUDIO_BASE_URL = _SPGISPEECH_BASE_URL + "audio"
 
 _SPGISPEECH_SUBSET_TO_DIR = {
     "s": ["s"],
@@ -1385,7 +1384,7 @@ _SPGISPEECH_AUDIO_ARCHIVES_NAMES = {
     "test": [f"test_part_{i}.tar.gz" for i in range(0, 3)],
 }
 
-_SPGISPEECH_META_BASE_URL = _SPGISPEECH_BASE_URL + "/meta"
+_SPGISPEECH_META_BASE_URL = _SPGISPEECH_BASE_URL + "meta"
 
 _SPGISPEECH_META_FILENAMES = {
     "s": "train_small.csv",
@@ -1417,6 +1416,8 @@ _LIBRISPEECH_DL_URLS = {
 
 _COMMON_VOICE_API_URL = "https://commonvoice.mozilla.org/api/v1"
 
+_COMMON_VOICE_BUNDLE_URL_TEMPLATE = 'cv-corpus-9.0-2022-04-27/cv-corpus-9.0-2022-04-27-{locale}.tar.gz'
+
 _TEDLIUM_BASE_URL = "https://huggingface.co/datasets/LIUM/tedlium/resolve/main/TEDLIUM_release3/legacy/"
 
 _TEDLIUM_URLS = {