From e4811360c81aaafd2d1f514969f87c096d4a39a3 Mon Sep 17 00:00:00 2001
From: masader-bot <zaidalyafeai@gmail.com>
Date: Sat, 6 Jun 2026 16:56:10 +0000
Subject: [PATCH] Creating datasets/mgb.json

---
 datasets/mgb.json | 62 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)
 create mode 100644 datasets/mgb.json

diff --git a/datasets/mgb.json b/datasets/mgb.json
new file mode 100644
index 00000000..44552188
--- /dev/null
+++ b/datasets/mgb.json
@@ -0,0 +1,62 @@
+{
+    "Name": "MGB",
+    "Volume": 1232.0,
+    "Unit": "hours",
+    "License": "unknown",
+    "Link": "https://arabicspeech.org/",
+    "HF_Link": "",
+    "Year": 2021,
+    "Domain": [
+        "news articles",
+        "TV channels"
+    ],
+    "Form": "audio",
+    "Collection_Style": [
+        "crawling",
+        "human annotation",
+        "manual curation"
+    ],
+    "Description": "Benchmark for Arabic ASR systems.",
+    "Ethical_Risks": "Medium",
+    "Provider": [
+        "HBKU",
+        "Qatar Computing Research Institute",
+        "Kanari AI",
+        "Carnegie Mellon University"
+    ],
+    "Derived_From": [
+        "MGB2",
+        "MGB3",
+        "MGB5"
+    ],
+    "Paper_Title": "Arabic Speech Recognition by End-to-End, Modular Systems and Human",
+    "Paper_Link": "https://arxiv.org/pdf/2101.08454v2",
+    "Tokenized": false,
+    "Host": "other",
+    "Access": "Free",
+    "Cost": "",
+    "Test_Split": true,
+    "Tasks": [
+        "speech recognition"
+    ],
+    "Venue_Title": "Journal of Computer Speech and Language",
+    "Venue_Type": "journal",
+    "Venue_Name": "Journal of Computer Speech and Language",
+    "Authors": [
+        "Amir Hussein",
+        "Shinji Watanabe",
+        "Ahmed Ali"
+    ],
+    "Affiliations": [
+        "HBKU",
+        "Qatar Computing Research Institute",
+        "Kanari AI",
+        "Carnegie Mellon University"
+    ],
+    "Abstract": "Recent advances in automatic speech recognition (ASR) have achieved accuracy levels comparable to human transcribers, which led researchers to debate if the machine has reached human performance. Previous work focused on the English language and modular hidden Markov model-deep neural network (HMM-DNN) systems. In this paper, we perform a comprehensive benchmarking for end-to-end transformer ASR, modular HMM-DNN ASR, and human speech recognition (HSR) on the Arabic language and its dialects. For the HSR, we evaluate linguist performance and lay-native speaker performance on a new dataset collected as a part of this study. For ASR the end-to-end work led to 12.5%, 27.5% , 33.8% WER; a new performance milestone for the MGB2, MGB3, and MGB5 challenges respectively. Our results suggest that human performance in the Arabic language is still considerably better than the machine with an absolute WER gap of 3.5% on average.",
+    "Subsets": [],
+    "Dialect": "mixed",
+    "Language": "ar",
+    "Script": "Arab",
+    "Added_By": "qwen/qwen3.6-35b-a3b"
+}
\ No newline at end of file