Upload dataset scripts

accents.py

@@ -0,0 +1,4 @@
+ACCENTS = {
+    "ar": "arabic",
+    "zh": "chinese"
+}

generate_train_test_split.py

@@ -0,0 +1,108 @@
+# This file is used to generate the train and test split of files within
+# the file system.
+
+import os
+import re
+import shutil
+import tarfile
+from math import floor
+from random import shuffle
+
+
+def get_file_list_from_dir(data_dir):
+    """Get a list of files within the directory"""
+    all_files = os.listdir(os.path.abspath(data_dir))
+    data_files = list(filter(lambda file: file.endswith('.wav'), all_files))
+    return data_files
+
+
+# Due to the audio and transcript files are in different folders, we need to split
+# them into train and test splits while maintaining their correct mapping in both splits
+# For example, if transcript 0001 is in test split, audio 0001 should also be in test split.
+#
+# Hence, instead of splitting the files itself, I will split the files based on their
+# ID number. Because every transcript and audio file has an ID number.
+# For example, arctic_a0001.txt is the transcript for arctic_a0001.wav.
+#
+# After splitting the ID numbers, we will then assign the files to their respective splits
+# based on their ID numbers.
+def shuffle_file_id(files):
+    file_id_regex = re.compile(r'[a-z]\d\d\d\d')
+    file_id = []
+
+    for file in files:
+        file_id.append(file_id_regex.search(file).group())
+
+    shuffle(file_id)
+    return file_id
+
+
+def train_test_splits(file_ids, training_ratio):
+    split=training_ratio
+    split_index = floor(len(file_ids) * split)
+    train = file_ids[:split_index]
+    test = file_ids[split_index:]
+    return train, test
+
+
+
+if __name__ == "__main__":
+    print("Running")
+    audio_file_paths = {
+        "ABA": "C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\ABA\\ABA\\wav",
+        "BWC": "C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\BWC\\BWC\\wav"
+    }
+    speaker_accent_map = {
+        "ABA": "ar",
+        "BWC": "zh"
+    }
+    splits = ["train", "test"]
+
+    try:
+        for speaker, accent in speaker_accent_map.items():
+            audio_files = get_file_list_from_dir(audio_file_paths[speaker])
+            shuffled_file_id = shuffle_file_id(audio_files)
+            train, test = train_test_splits(shuffled_file_id, 0.7)
+
+            print(train)
+            print("Number of samples in training: ", len(train))
+            print(test)
+            print("Number of samples in testing: ", len(test))
+            tar_file = tarfile.open(
+                f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\audio\\{accent}\\train\\{accent}_train.tar",
+                "w"
+            )
+            for id in train:
+                # Copy training audio into a tar file
+                tar_file.add(
+                    f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\wav\\arctic_{id}.wav",
+                    arcname=f"arctic_{id}.wav"
+                )
+
+                # Copy training transcript into a tsv file
+                with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\transcript\\{accent}\\train.tsv", 'a') as tsv_file:
+                    with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\transcript\\arctic_{id}.txt") as txt_file:
+                        tsv_file.write(txt_file.read() + "\n")
+                tsv_file.close()
+            tar_file.close()
+
+            # Copy testing audio into tar file
+            tar_file = tarfile.open(
+                f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\audio\\{accent}\\test\\{accent}_test.tar",
+                "w"
+            )
+            for id in test:
+                tar_file.add(
+                    f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\wav\\arctic_{id}.wav",
+                    arcname=f"arctic_{id}.wav"
+                )
+
+                # Copy testing transcript into tsv file
+                with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\transcript\\{accent}\\test.tsv", 'a') as tsv_file:
+                    with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\transcript\\arctic_{id}.txt") as txt_file:
+                        tsv_file.write(txt_file.read() + "\n")
+                tsv_file.close()
+            tar_file.close()
+
+    except PermissionError:
+        print("Permission denied")

release_stats.py

@@ -0,0 +1,8 @@
+STATS = {
+    "name": "VerbaLex Voice",
+    "version": "1.0.0",
+    "accents": {
+        "ar": {"numOfSpeaker": "1", "numOfWavFiles": "1129"},
+        "zh": {"numOfSpeaker": "1", "numOfWavFiles": "1130"}
+    }
+}

verbalex_voice.py

@@ -0,0 +1,134 @@
+import csv
+import os
+
+import datasets
+from tqdm import tqdm
+
+from VerbaLex_Voice.accents import ACCENTS
+from VerbaLex_Voice.release_stats import STATS
+
+_HOMEPAGE = "https://huggingface.co/datasets/RitchieP/VerbaLex_voice"
+
+_LICENSE = "https://choosealicense.com/licenses/apache-2.0/"
+
+_BASE_URL = "https://huggingface.co/datasets/RitchieP/VerbaLex_voice/tree/main"
+
+_AUDIO_URL = _BASE_URL + "audio/{accent}/{split}/{accent}_{split}.tar"
+
+_TRANSCRIPT_URL = _BASE_URL + "transcript/{accent}/{split}.tsv"
+
+_CITATION = """\
+"""
+
+
+class VerbaLexVoiceConfig(datasets.BuilderConfig):
+    def __init__(self, name, version, **kwargs):
+        self.accent = kwargs.pop("accent", None)
+        self.num_speakers = kwargs.pop("num_speakers", None)
+        self.num_files = kwargs.pop("num_clips", None)
+        description = (
+            f"VerbaLex Voice english speech-to-text dataset in {self.accent} accent."
+        )
+
+        super(VerbaLexVoiceConfig, self).__init__(
+            name=name,
+            version=datasets.Version(version),
+            description=description,
+            **kwargs,
+        )
+
+
+class VerbaLexVoiceDataset(datasets.GeneratorBasedBuilder):
+    """
+    VerbaLex is a dataset containing different English accents from non-native English speakers.
+    This dataset is created directly from the L2-Arctic dataset.
+    """
+    BUILDER_CONFIGS = [
+        VerbaLexVoiceConfig(
+            name=accent,
+            version=STATS["version"],
+            accent=ACCENTS[accent],
+            num_speakers=accent_stats["numOfSpeaker"],
+            num_files=accent_stats["numOfWavFiles"]
+        )
+        for accent, accent_stats in STATS["accents"].items()
+    ]
+
+    DEFAULT_CONFIG_NAME = "all"
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=(
+                "VerbaLex Voice is a speech dataset focusing on accented English speech."
+                "It specifically targets speeches from speakers that is a non-native English speaker."
+            ),
+            features=datasets.Features(
+                {
+                    "path": datasets.Value("string"),
+                    "accent": datasets.Value("string"),
+                    "sentence": datasets.Value("string"),
+                    "audio": datasets.Audio(sampling_rate=44_100)
+                }
+            ),
+            supervised_keys=None,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION
+        )
+
+    def _split_generators(self, dl_manager):
+        """Returns SplitGenerators"""
+        accent = self.config.name
+
+        splits = ("train", "test")
+        audio_urls = {}
+        for split in splits:
+            audio_urls[split] = _AUDIO_URL.format(accent=accent, split=split)
+        archive_paths = dl_manager.download(audio_urls)
+        local_extracted_archive_paths = dl_manager.extract(archive_paths) if not dl_manager.is_streaming else {}
+
+        meta_urls = {split: _TRANSCRIPT_URL.format(accent=accent, split=split) for split in splits}
+        meta_paths = dl_manager.download_and_extract(meta_urls)
+
+        split_names = {
+            "train": datasets.Split.TRAIN,
+            "test": datasets.Split.TEST
+        }
+        split_generators = []
+        for split in splits:
+            split_generators.append(
+                datasets.SplitGenerator(
+                    name=split_names.get(split, split),
+                    gen_kwargs={
+                        "local_extracted_archive_paths": local_extracted_archive_paths.get(split),
+                        "archives": [dl_manager.iter_archive(path) for path in archive_paths.get(split)],
+                        "meta_path": meta_paths[split]
+                    }
+                )
+            )
+
+        return split_generators
+
+    def _generate_examples(self, local_extracted_archive_paths, archives, meta_path):
+        data_fields = list(self._info().features.keys())
+        metadata = {}
+        with open(meta_path, encoding="UTF-8") as f:
+            reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
+            for row in tqdm(reader, desc="Reading metadata..."):
+                if not row["path"].endswith(".wav"):
+                    row["path"] += ".wav"
+                for field in data_fields:
+                    if field not in row:
+                        row[field] = ""
+                metadata[row["path"]] = row
+
+        for i, audio_archive in enumerate(archives):
+            for path, file in audio_archive:
+                _, filename = os.path.split(path)
+                if filename in metadata:
+                    result = dict(metadata[filename])
+                    path = os.path.join(local_extracted_archive_paths[i],
+                                        path) if local_extracted_archive_paths else path
+                    result["audio"] = {"path": path, "bytes": file.read()}
+                    result["path"] = path
+                    yield path, result