Initial commit

.gitattributes

@@ -35,3 +35,9 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.mp3 filter=lfs diff=lfs merge=lfs -text
 *.ogg filter=lfs diff=lfs merge=lfs -text
 *.wav filter=lfs diff=lfs merge=lfs -text
+dataset/lexicon_test.json filter=lfs diff=lfs merge=lfs -text
+dataset/lexicon_train.json filter=lfs diff=lfs merge=lfs -text
+dataset/lexicon_valid.json filter=lfs diff=lfs merge=lfs -text
+dataset/sentence_test.json filter=lfs diff=lfs merge=lfs -text
+dataset/sentence_train.json filter=lfs diff=lfs merge=lfs -text
+dataset/sentence_valid.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,9 @@
+# librig2p-nostress - Grapheme-To-Phoneme Dataset
+
+This dataset contains samples that can be used to train a Grapheme-to-Phoneme system **without** stress information.
+
+The dataset is derived from the following pre-existing datasets:
+
+* [LibriSpeech ASR Corpus](https://www.openslr.org/12)
+* [LibriSpeech Alignments](https://github.com/CorentinJ/librispeech-alignments)
+* [Wikipedia Homograph Disambiguation Data](https://github.com/google/WikipediaHomographData)

dataset/lexicon_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0347606ab149f930db118d647fffd70dfd044fa9e25939b123f06c6aa8de51b4
+size 361442

dataset/lexicon_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d36f6af961d981788a9279c2739bac17cf6e00c01ca2cff35d0b90b0b87d1f6c
+size 35212147

dataset/lexicon_valid.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b422797758d6c020fc3df607ca72adb41f5ac252024fe351e216db4c8387322
+size 358580

dataset/sentence_test.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18fe6d8d3202c312aa1916b7c3b4d908e5ee2acb295b3b1309fd5ac2f4841125
+size 3318940

dataset/sentence_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:493914a5e9a4114b2e90736d2969aa4bc99bcd5898eec2070d9aa51a962712f3
+size 217148647

dataset/sentence_valid.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c6600d164d75ddffd68f62f572aec98660b06b0fef35bfb6226afdd8331855f
+size 3406399

librig2p-nostress-space-cmu.py

@@ -0,0 +1,84 @@
+# coding=utf-8
+# Copyright 2021 Artem Ploujnikov
+
+
+# Lint as: python3
+import json
+
+import datasets
+
+_DESCRIPTION = """\
+Grapheme-to-Phoneme training, validation and test sets
+"""
+
+_BASE_URL = "https://huggingface.co/datasets/flexthink/librig2p-nostress-space/resolve/main/dataset"
+_HOMEPAGE_URL = "https://huggingface.co/datasets/flexthink/librig2p-nostress-space"
+_NA = "N/A"
+_SPLIT_TYPES = ["train", "valid", "test"]
+_DATA_TYPES = ["lexicon", "sentence", "homograph"]
+_SPLITS = [
+    f"{data_type}_{split_type}"
+    for data_type in _DATA_TYPES
+    for split_type in _SPLIT_TYPES
+]
+
+
+class GraphemeToPhoneme(datasets.GeneratorBasedBuilder):
+    def __init__(self, base_url=None, splits=None, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.base_url = base_url or _BASE_URL
+        self.splits = splits or _SPLITS
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features(
+                {
+                    "sample_id": datasets.Value("string"),
+                    "speaker_id": datasets.Value("string"),
+                    "origin": datasets.Value("string"),
+                    "char": datasets.Value("string"),
+                    "phn": datasets.Sequence(datasets.Value("string")),
+                    "homograph": datasets.Value("string"),
+                    "homograph_wordid": datasets.Value("string"),
+                    "homograph_char_start": datasets.Value("int32"),
+                    "homograph_char_end": datasets.Value("int32"),
+                    "homograph_phn_start": datasets.Value("int32"),
+                    "homograph_phn_end": datasets.Value("int32"),
+                },
+            ),
+            supervised_keys=None,
+            homepage=_HOMEPAGE_URL,
+        )
+
+    def _get_url(self, split):
+        return f"{self.base_url}/{split}.json"
+
+    def _split_generator(self, dl_manager, split):
+        url = self._get_url(split)
+        path = dl_manager.download_and_extract(url)
+        return datasets.SplitGenerator(
+            name=split, gen_kwargs={"datapath": path, "datatype": split},
+        )
+
+    def _split_generators(self, dl_manager):
+        return [self._split_generator(dl_manager, split) for split in self.splits]
+
+    def _generate_examples(self, datapath, datatype):
+        with open(datapath, encoding="utf-8") as f:
+            data = json.load(f)
+        for sentence_counter, (sample_id, item) in enumerate(data.items()):
+            resp = {
+                "sample_id": sample_id,
+                "speaker_id": str(item.get("speaker_id") or _NA),
+                "origin": item["origin"],
+                "char": item["char"],
+                "phn": item["phn"],
+                "homograph": item.get("homograph", _NA),
+                "homograph_wordid": item.get("homograph_wordid", _NA),
+                "homograph_char_start": item.get("homograph_char_start", 0),
+                "homograph_char_end": item.get("homograph_char_end", 0),
+                "homograph_phn_start": item.get("homograph_phn_start", 0),
+                "homograph_phn_end": item.get("homograph_phn_end", 0),
+            }
+            yield sentence_counter, resp