إضافة التوثيق
Browse files
README.md
CHANGED
|
@@ -67,3 +67,157 @@ configs:
|
|
| 67 |
- split: train
|
| 68 |
path: data/train-*
|
| 69 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
- split: train
|
| 68 |
path: data/train-*
|
| 69 |
---
|
| 70 |
+
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
* قاعدة بيانات معيارية صغيرة الحجم (159 عينة) لأجل تقييم أداء نماذج الذكاء الاصطناعي لكشف أخطاء التلاوة وقواعد تجويد القرآن الكريم.
|
| 75 |
+
* لقد تم بناء هذه قاعدة البيانات على [qdat](https://huggingface.co/datasets/obadx/qdat) من [الورقة البحثية](https://www.researchgate.net/profile/Ban-Sharief/publication/350785609_QDAT_A_data_set_for_Reciting_the_Quran/links/6071a3a192851c8a7bb7afd2/QDAT-A-data-set-for-Reciting-the-Quran.pdf)
|
| 76 |
+
* المقاطع عبارة عن قارءة لآية: {قَالُوا۟ لَا عِلْمَ لَنَآ إِنَّكَ أَنتَ عَلَّٰمُ ٱلْغُيُوبِ} بسورة المائدة من الآية 109
|
| 77 |
+
|
| 78 |
+
# آلية بناء البيانات المعيارية
|
| 79 |
+
كان لدى qdat بعض العيوب مما أدى لبناء qdat_bench أهمها:
|
| 80 |
+
* عدم تغطية كل قواعد التجويد
|
| 81 |
+
* كانت تحتوي على 161 قارئ وقام كل قارئ بتسجيل 10 مقاطع لنفس الأحكام والقواعد فقمنا باختيار مقطع لك قارئ
|
| 82 |
+
|
| 83 |
+
## هيكل البيانات
|
| 84 |
+
|
| 85 |
+
### الميزات الرئيسية:
|
| 86 |
+
|
| 87 |
+
- **audio**: الملف الصوتي (معدل العينة: غير محدد، أحادي القناة)
|
| 88 |
+
- **id**: معرف فريد لكل عنصر
|
| 89 |
+
- **original_id**: المعرف الأصلي في مجموعة البيانات الأصلية
|
| 90 |
+
- **gender**: جنس القارئ (ذكر/أنثى)
|
| 91 |
+
- **age**: عمر القارئ
|
| 92 |
+
- **phonetic_transcript**: النص الصوتي باستخدام حزمة [quran-transcript](https://github.com/obadx/quran-transcript)
|
| 93 |
+
|
| 94 |
+
### أحكام المدود:
|
| 95 |
+
|
| 96 |
+
- `qalo_alif_len`: طول مد الألف في كلمة "قالوا" (0-8)
|
| 97 |
+
- `qalo_waw_len`: طول مد الواو في كلمة "قالوا" (0-8)
|
| 98 |
+
- `laa_alif_len`: طول مد الألف في كلمة "لا" (0-8)
|
| 99 |
+
- `separate_madd`: طول المد المنفصل في "لنا إنك" (0-8)
|
| 100 |
+
- `allam_alif_len`: طول مد الألف في كلمة "علام" (0-8)
|
| 101 |
+
- `madd_aared_len`: طول المد العارض للسكون (0-8)
|
| 102 |
+
|
| 103 |
+
### أحكام الغنة:
|
| 104 |
+
|
| 105 |
+
- `noon_moshaddadah_len`: طول النون المشددة في "إنَّك" (0=جزئي, 1=كامل)
|
| 106 |
+
- `noon_mokhfah_len`: طول النون المخفاة في "أنت" (0=نون, 1=جزئي, 2=كامل)
|
| 107 |
+
|
| 108 |
+
### أحكام القلقلة:
|
| 109 |
+
|
| 110 |
+
- `qalqalah`: وجود القلقلة في "الغيوب" (0=لا يوجد, 1=يوجد)
|
| 111 |
+
|
| 112 |
+
### صفات الحروف:
|
| 113 |
+
|
| 114 |
+
كل مجموعة حروف صوتية مثل: "اااا" في ال `phonetic_transcript` لها 10 صفات في العمود `sifat` على هيئة dictionary
|
| 115 |
+
|
| 116 |
+
- `ghonna`: الغنة
|
| 117 |
+
- `hams_or_jahr`: الهمس أو الجهر
|
| 118 |
+
- `istitala`: الاستطالة
|
| 119 |
+
- `itbaq`: الإطباق
|
| 120 |
+
- `phonemes`: الأصوات
|
| 121 |
+
- `qalqla`: القلقلة
|
| 122 |
+
- `safeer`: الصفير
|
| 123 |
+
- `shidda_or_rakhawa`: الشدة أو الرخاوة
|
| 124 |
+
- `tafashie`: التفشي
|
| 125 |
+
- `tafkheem_or_taqeeq`: التفخيم أو الترقيق
|
| 126 |
+
- `tikraar`: التكرار
|
| 127 |
+
|
| 128 |
+
## الاستخدام
|
| 129 |
+
|
| 130 |
+
```python
|
| 131 |
+
from datasets import load_dataset
|
| 132 |
+
|
| 133 |
+
ds = load_dataset('obadx/qdat_bench')
|
| 134 |
+
print(ds['train'][0]) # عرض العينة الأولى
|
| 135 |
+
```
|
| 136 |
+
|
| 137 |
+
## الإحصائيات
|
| 138 |
+
|
| 139 |
+
- عدد العينات: 159
|
| 140 |
+
- Split: train فقط
|
| 141 |
+
|
| 142 |
+

|
| 143 |
+
|
| 144 |
+

|
| 145 |
+
|
| 146 |
+

|
| 147 |
+
|
| 148 |
+
|
| 149 |
+
# English
|
| 150 |
+
|
| 151 |
+
# QDat-Bench Dataset
|
| 152 |
+
|
| 153 |
+
A benchmark dataset for evaluating model performance in processing Quranic audio recordings with focus on Tajweed rules.
|
| 154 |
+
|
| 155 |
+
## Description
|
| 156 |
+
|
| 157 |
+
This dataset contains Quranic audio recordings with detailed annotations for Tajweed characteristics. It is designed to serve as a benchmark for evaluating audio processing models in recognizing Tajweed rules.
|
| 158 |
+
|
| 159 |
+
## Data Structure
|
| 160 |
+
|
| 161 |
+
### Main Features:
|
| 162 |
+
|
| 163 |
+
- **audio**: Audio file (sampling rate: None, mono channel)
|
| 164 |
+
- **id**: Unique identifier for each element
|
| 165 |
+
- **original_id**: The item's ID in the original dataset
|
| 166 |
+
- **gender**: Reciter's gender (male/female)
|
| 167 |
+
- **age**: Reciter's age
|
| 168 |
+
- **phonetic_transcript**: Phonetic transcription using `quran-transcript` package
|
| 169 |
+
|
| 170 |
+
### Madd (Prolongation) Rules:
|
| 171 |
+
|
| 172 |
+
- `qalo_alif_len`: Length of normal madd alif in word "قالوا" (0-8)
|
| 173 |
+
- `qalo_waw_len`: Length of normal madd waw in word "قالوا" (0-8)
|
| 174 |
+
- `laa_alif_len`: Length of normal madd alif in word "لا" (0-8)
|
| 175 |
+
- `separate_madd`: Length of separate madd for "لنا إنك" (0-8)
|
| 176 |
+
- `allam_alif_len`: Length of normal madd alif in word "علام" (0-8)
|
| 177 |
+
- `madd_aared_len`: Length of madd aared for sukoon (0-8)
|
| 178 |
+
|
| 179 |
+
### Ghunnah (Nasalization) Rules:
|
| 180 |
+
|
| 181 |
+
- `noon_moshaddadah_len`: Length of noon moshaddadah in "إنَّك" (0=partial, 1=complete)
|
| 182 |
+
- `noon_mokhfah_len`: Length of noon mokhfah in "أنت" (0=noon, 1=partial, 2=complete)
|
| 183 |
+
|
| 184 |
+
### Qalqalah (Echo) Rules:
|
| 185 |
+
|
| 186 |
+
- `qalqalah`: Existence of qalqalah for "الغيوب" (0=no qalqalah, 1=has qalqalah)
|
| 187 |
+
|
| 188 |
+
### Letter Characteristics:
|
| 189 |
+
|
| 190 |
+
Contains a list of characteristics for each letter:
|
| 191 |
+
|
| 192 |
+
- `ghonna`: Ghunnah (nasalization)
|
| 193 |
+
- `hams_or_jahr`: Hams (whisper) or Jahr (clarity)
|
| 194 |
+
- `istitala`: Istitala (elongation)
|
| 195 |
+
- `itbaq`: Itbaq (adhesion)
|
| 196 |
+
- `phonemes`: Phonemes
|
| 197 |
+
- `qalqla`: Qalqalah (echo)
|
| 198 |
+
- `safeer`: Safeer (whistling)
|
| 199 |
+
- `shidda_or_rakhawa`: Shidda (strength) or Rakhawa (softness)
|
| 200 |
+
- `tafashie`: Tafashie (diffusion)
|
| 201 |
+
- `tafkheem_or_taqeeq`: Tafkheem (emphasis) or Taqeeq (thinning)
|
| 202 |
+
- `tikraar`: Tikraar (repetition)
|
| 203 |
+
|
| 204 |
+
## Usage
|
| 205 |
+
|
| 206 |
+
```python
|
| 207 |
+
from datasets import load_dataset
|
| 208 |
+
|
| 209 |
+
ds = load_dataset('obadx/qdat_bench')
|
| 210 |
+
print(ds['train'][0]) # Display first sample
|
| 211 |
+
```
|
| 212 |
+
|
| 213 |
+
## Statistics
|
| 214 |
+
|
| 215 |
+
- Number of samples: 159
|
| 216 |
+
- Split: train only
|
| 217 |
+
|
| 218 |
+
|
| 219 |
+

|
| 220 |
+
|
| 221 |
+

|
| 222 |
+
|
| 223 |
+

|