Upload README.md with huggingface_hub

README.md

@@ -9,35 +9,18 @@ tags:
 - phonemization
 - IPA
 - wikipedia
-pretty_name: Basque Wikipedia Phonemized Corpus (IPA)
+pretty_name: Basque Wikipedia Phonemized Corpus (Text + IPA phonemes)
 size_categories:
 - 1M<n<10M
-dataset_info:
-  features:
-  - name: text
-    dtype: string
-  - name: phonemes
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 653374158
-    num_examples: 1672981
-  download_size: 300920769
-  dataset_size: 653374158
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
 ---
 
-# Basque Wikipedia Phonemized Corpus (IPA)
+# Basque Wikipedia Phonemized Corpus (Text + IPA phonemes)
 
 ## Dataset Description
 
-A large-scale phonemized corpus derived from the Basque Wikipedia dump. Each row contains a paragraph-level sequence of standard IPA phonemes, with stressed vowels represented using the apostrophe convention (e.g. `'a`, `'e`, `'i`, `'o`, `'u`) and affricates as multicharacter sequences (e.g. `tʃ`, `tʂ`, `ts`).
+A large-scale paired corpus derived from the Basque Wikipedia dump. Each row contains both the **original plain text** and its **IPA phoneme transcription**, at paragraph level. Stressed vowels use the apostrophe convention (e.g. `'a`, `'e`, `'i`, `'o`, `'u`) and affricates are kept as multicharacter sequences (e.g. `tʃ`, `tʂ`, `ts`).
 
-This dataset is intended for training phonetic encoders (PL-BERT), text-to-speech (TTS) and speech-related language models for Basque.
+This dataset is intended for training text-to-speech (TTS) and grapheme-to-phoneme (G2P) models for Basque.
 
 ---
 
@@ -45,7 +28,7 @@ This dataset is intended for training phonetic encoders (PL-BERT), text-to-speec
 
 | Split | Rows |
 |-------|------|
-| train | 1,672,981 |
+| samples | 1,672,981 |
 
 ---
 
@@ -53,25 +36,19 @@ This dataset is intended for training phonetic encoders (PL-BERT), text-to-speec
 
 ### Fields
 
-- `phonemes` (`string`): A space-separated sequence of IPA phonemes at word level. Punctuation marks are attached directly to the preceding word (e.g. `astrOnomia.` not `astrOnomia .`).
+- `text` (`string`): The Basque Wikipedia plain text, at paragraph level.
+- `phonemes` (`string`): The corresponding IPA phoneme transcription. Words are space-separated; punctuation is attached directly to the preceding word (e.g. `astr'onomia.` not `astr'onomia .`).
 
 ### Example
 
 ```python
 from datasets import load_dataset
 
-ds = load_dataset("<your-username>/basque-wiki-phonemes", split="train")
+ds = load_dataset("HiTZ/wikipedia_basque_ipa", split="train")
+print(ds[0]["text"])
+# → "Historiako lehenengo zientzia izan da astronomia. Zibilizazio eta kultura guztiek..."
 print(ds[0]["phonemes"])
-# → "istoɾiako le'enenɡo ʂi'entʂia iʂ'an da astr'onomia. ʂiβ'iliʂaʂio eta kult'uɾa ..."
-```
-
-**Split into individual phoneme tokens:**
-
-```python
-for example in ds.select(range(5)):
-    tokens = example["phonemes"].split()
-    print(tokens)
-    # e.g. ["istoɾiako", "le'enenɡo", "ʂi'entʂia", "iʂ'an", "da", "astr'onomia.", ...]
+# → "'istoɾiako le'enenɡo ʂi'entʂia iʂ'an da astr'onomia. ʂiβ'iliʂaʂio eta kult'uɾa..."
 ```
 
 ---
@@ -82,12 +59,13 @@ This dataset uses **standard IPA** symbols.
 
 ---
 
-## Data Pipeline
+## Data Processing Pipeline
 
-### Step 1 — Wikipedia Extraction
+Data processing steps from raw data extracted from Wikipedia to the phonemized utterances.
 
-Raw text was extracted from the Basque Wikipedia XML dump using a customized version of [WikiExtractor](https://github.com/attardi/wikiextractor). The extractor:
+### Step 1 — Wikipedia Extraction (`WikiExtractor.py`)
 
+Raw text was extracted from the Basque Wikipedia XML dump using a customized version of [WikiExtractor](https://github.com/attardi/wikiextractor). The extractor:
 - Strips MediaWiki markup (templates, tables, infoboxes)
 - Expands wiki links to their anchor text
 - Removes HTML tags, comments, and special elements (`<ref>`, `<math>`, etc.)
@@ -99,7 +77,6 @@ Raw text was extracted from the Basque Wikipedia XML dump using a customized ver
 The extracted text was further cleaned with the following filters and transformations:
 
 **Sentence-level filters (removal):**
-
 - Sentences shorter than **100 characters**
 - Sentences containing **double quotes** (`"`)
 - Sentences containing **chess notation** (`e4`, `c4`)
@@ -110,7 +87,6 @@ The extracted text was further cleaned with the following filters and transforma
 - Sentences consisting **only of digits and punctuation**
 
 **Text normalization transformations:**
-
 - `"K. a."` → `"K.a."` and `"K. o."` → `"K.o."` (Basque grammatical abbreviations)
 - Dots after single uppercase initials removed (e.g. `A.` → `A`)
 - Content inside parentheses/brackets removed
@@ -139,6 +115,7 @@ Each cleaned sentence was first **normalized** and then **phonemized** using **[
 - Outputs IPA symbols with stress markers and multicharacter affricates preserved
 - Attaches punctuation marks to the preceding word
 
+
 ---
 
 ## Usage
@@ -146,10 +123,31 @@ Each cleaned sentence was first **normalized** and then **phonemized** using **[
 ```python
 from datasets import load_dataset
 
-ds = load_dataset("<your-username>/basque-wiki-phonemes", split="train")
+ds = load_dataset("HiTZ/wikipedia_basque_ipa", split="train")
 
 for example in ds.select(range(5)):
+    print(example["text"])
     print(example["phonemes"])
+    print()
+```
+
+**Split phonemes into individual word tokens:**
+
+```python
+for example in ds.select(range(5)):
+    tokens = example["phonemes"].split()
+    print(tokens)
+    # e.g. ["'istoɾiako", "le'enenɡo", "ʂi'entʂia", "iʂ'an", "da", "astr'onomia.", ...]
+```
+
+**Use as a G2P training corpus:**
+
+```python
+for example in ds.select(range(5)):
+    words = example["text"].split()
+    phonemes = example["phonemes"].split()
+    # Note: words and phoneme tokens are aligned one-to-one
+    # (punctuation is attached to the preceding phoneme token)
 ```
 
 ---
@@ -162,10 +160,9 @@ The dataset is derived from Basque Wikipedia, which is released under the [Creat
 
 ## Citation
 
-If you use this dataset, please cite the Basque Wikipedia and acknowledge the phonemization pipeline developed at HiTZ/AhoLab (University of the Basque Country).
-
----
+If you use this dataset, please cite the Basque Wikipedia and acknowledge the phonemization pipeline developed at AhoLab (University of the Basque Country).
 
 ## Related Resources
 
 - **[ahoNT](https://github.com/hitz-zentroa/ahoNT)** — Basque text normalization and phonemization tool
+- **[WikiExtractor](https://github.com/attardi/wikiextractor)** - Python script that extracts and cleans text from a Wikipedia database backup dump