diff --git a/mteb/descriptive_stats/AudioClassification/GLOBEV3Age.json b/mteb/descriptive_stats/AudioClassification/GLOBEV3Age.json
new file mode 100644
index 0000000000..325762115f
--- /dev/null
+++ b/mteb/descriptive_stats/AudioClassification/GLOBEV3Age.json
@@ -0,0 +1,106 @@
+{
+    "test": {
+        "num_samples": 5000,
+        "number_texts_intersect_with_train": null,
+        "text_statistics": null,
+        "image_statistics": null,
+        "audio_statistics": {
+            "total_duration_seconds": 25920.202374999975,
+            "min_duration_seconds": 0.56,
+            "average_duration_seconds": 5.184040474999995,
+            "max_duration_seconds": 22.78,
+            "unique_audios": 5000,
+            "average_sampling_rate": 24000.0,
+            "sampling_rates": {
+                "24000": 5000
+            }
+        },
+        "label_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.0,
+            "max_labels_per_text": 1,
+            "unique_labels": 9,
+            "labels": {
+                "twenties": {
+                    "count": 1378
+                },
+                "fifties": {
+                    "count": 638
+                },
+                "teens": {
+                    "count": 606
+                },
+                "fourties": {
+                    "count": 570
+                },
+                "thirties": {
+                    "count": 1382
+                },
+                "sixties": {
+                    "count": 263
+                },
+                "seventies": {
+                    "count": 160
+                },
+                "nineties": {
+                    "count": 2
+                },
+                "eighties": {
+                    "count": 1
+                }
+            }
+        }
+    },
+    "train": {
+        "num_samples": 5000,
+        "number_texts_intersect_with_train": null,
+        "text_statistics": null,
+        "image_statistics": null,
+        "audio_statistics": {
+            "total_duration_seconds": 25967.226791666697,
+            "min_duration_seconds": 1.08,
+            "average_duration_seconds": 5.1934453583333395,
+            "max_duration_seconds": 10.539958333333333,
+            "unique_audios": 5000,
+            "average_sampling_rate": 24000.0,
+            "sampling_rates": {
+                "24000": 5000
+            }
+        },
+        "label_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.0,
+            "max_labels_per_text": 1,
+            "unique_labels": 9,
+            "labels": {
+                "fourties": {
+                    "count": 617
+                },
+                "thirties": {
+                    "count": 617
+                },
+                "twenties": {
+                    "count": 617
+                },
+                "teens": {
+                    "count": 617
+                },
+                "fifties": {
+                    "count": 618
+                },
+                "sixties": {
+                    "count": 618
+                },
+                "seventies": {
+                    "count": 618
+                },
+                "eighties": {
+                    "count": 618
+                },
+                "nineties": {
+                    "count": 60
+                }
+            }
+        }
+    }
+}
\ No newline at end of file
diff --git a/mteb/descriptive_stats/AudioClassification/GLOBEV3Gender.json b/mteb/descriptive_stats/AudioClassification/GLOBEV3Gender.json
new file mode 100644
index 0000000000..d56b9e88f8
--- /dev/null
+++ b/mteb/descriptive_stats/AudioClassification/GLOBEV3Gender.json
@@ -0,0 +1,64 @@
+{
+    "test": {
+        "num_samples": 5000,
+        "number_texts_intersect_with_train": null,
+        "text_statistics": null,
+        "image_statistics": null,
+        "audio_statistics": {
+            "total_duration_seconds": 25873.349833333417,
+            "min_duration_seconds": 0.56,
+            "average_duration_seconds": 5.174669966666683,
+            "max_duration_seconds": 21.984,
+            "unique_audios": 5000,
+            "average_sampling_rate": 24000.0,
+            "sampling_rates": {
+                "24000": 5000
+            }
+        },
+        "label_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.0,
+            "max_labels_per_text": 1,
+            "unique_labels": 2,
+            "labels": {
+                "1": {
+                    "count": 3275
+                },
+                "0": {
+                    "count": 1725
+                }
+            }
+        }
+    },
+    "train": {
+        "num_samples": 5000,
+        "number_texts_intersect_with_train": null,
+        "text_statistics": null,
+        "image_statistics": null,
+        "audio_statistics": {
+            "total_duration_seconds": 25205.078791666667,
+            "min_duration_seconds": 0.96,
+            "average_duration_seconds": 5.041015758333334,
+            "max_duration_seconds": 10.4,
+            "unique_audios": 5000,
+            "average_sampling_rate": 24000.0,
+            "sampling_rates": {
+                "24000": 5000
+            }
+        },
+        "label_statistics": {
+            "min_labels_per_text": 1,
+            "average_label_per_text": 1.0,
+            "max_labels_per_text": 1,
+            "unique_labels": 2,
+            "labels": {
+                "0": {
+                    "count": 2500
+                },
+                "1": {
+                    "count": 2500
+                }
+            }
+        }
+    }
+}
\ No newline at end of file
diff --git a/mteb/tasks/classification/eng/__init__.py b/mteb/tasks/classification/eng/__init__.py
index 9c99ebc387..8d31917eac 100644
--- a/mteb/tasks/classification/eng/__init__.py
+++ b/mteb/tasks/classification/eng/__init__.py
@@ -23,6 +23,12 @@
 from mteb.tasks.classification.eng.globe_v2_gender_classification import (
     GlobeV2GenderClassification,
 )
+from mteb.tasks.classification.eng.globe_v3_age_classification import (
+    GlobeV3AgeClassification,
+)
+from mteb.tasks.classification.eng.globe_v3_gender_classification import (
+    GlobeV3GenderClassification,
+)
 from mteb.tasks.classification.eng.iemocap_emotion import IEMOCAPEmotionClassification
 from mteb.tasks.classification.eng.iemocap_gender import IEMOCAPGenderClassification
 from mteb.tasks.classification.eng.libri_count import LibriCount
@@ -383,6 +389,8 @@
     "GTSRBClassification",
     "GlobeV2AgeClassification",
     "GlobeV2GenderClassification",
+    "GlobeV3AgeClassification",
+    "GlobeV3GenderClassification",
     "HUMEEmotionClassification",
     "HUMEToxicConversationsClassification",
     "HUMETweetSentimentExtractionClassification",
diff --git a/mteb/tasks/classification/eng/globe_v2_age_classification.py b/mteb/tasks/classification/eng/globe_v2_age_classification.py
index c8b09e6966..820fb4ab59 100644
--- a/mteb/tasks/classification/eng/globe_v2_age_classification.py
+++ b/mteb/tasks/classification/eng/globe_v2_age_classification.py
@@ -13,7 +13,7 @@ class GlobeV2AgeClassification(AbsTaskClassification):
         },
         type="AudioClassification",
         category="a2t",
-        eval_splits=["train"],
+        eval_splits=["test"],
         eval_langs=["eng-Latn"],
         main_score="accuracy",
         date=("2025-01-13", "2025-01-13"),
diff --git a/mteb/tasks/classification/eng/globe_v3_age_classification.py b/mteb/tasks/classification/eng/globe_v3_age_classification.py
new file mode 100644
index 0000000000..3450b653f3
--- /dev/null
+++ b/mteb/tasks/classification/eng/globe_v3_age_classification.py
@@ -0,0 +1,39 @@
+from mteb.abstasks.classification import AbsTaskClassification
+from mteb.abstasks.task_metadata import TaskMetadata
+
+
+class GlobeV3AgeClassification(AbsTaskClassification):
+    metadata = TaskMetadata(
+        name="GLOBEV3Age",
+        description="Age classification from the GLOBE v3 dataset (sampled and enhanced from CommonVoice dataset for TTS purpose). This dataset is a stratified and downsampled version of the original dataset, containing about 535 hours of speech data across 164 accents. We use the age column as the target label for audio classification.",
+        reference="https://huggingface.co/datasets/MushanW/GLOBE_V3",
+        dataset={
+            "path": "mteb/globe-v3-age-mini",
+            "revision": "f7399f4b836508a178c0913868e82462b4a8919b",
+        },
+        type="AudioClassification",
+        category="a2t",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="accuracy",
+        date=("2025-05-26", "2025-05-26"),
+        domains=["Spoken", "Speech"],
+        task_subtypes=["Age Classification"],
+        license="cc0-1.0",
+        annotations_creators="automatic",
+        dialect=[],
+        modalities=["audio"],
+        sample_creation="found",
+        bibtex_citation=r"""
+@misc{wang2024globe,
+  archiveprefix = {arXiv},
+  author = {Wenbin Wang and Yang Song and Sanjay Jha},
+  eprint = {2406.14875},
+  title = {GLOBE: A High-quality English Corpus with Global Accents for Zero-shot Speaker Adaptive Text-to-Speech},
+  year = {2024},
+}
+""",
+    )
+
+    input_column_name: str = "audio"
+    label_column_name: str = "predicted_age"
diff --git a/mteb/tasks/classification/eng/globe_v3_gender_classification.py b/mteb/tasks/classification/eng/globe_v3_gender_classification.py
new file mode 100644
index 0000000000..bf8f0cda07
--- /dev/null
+++ b/mteb/tasks/classification/eng/globe_v3_gender_classification.py
@@ -0,0 +1,39 @@
+from mteb.abstasks.classification import AbsTaskClassification
+from mteb.abstasks.task_metadata import TaskMetadata
+
+
+class GlobeV3GenderClassification(AbsTaskClassification):
+    metadata = TaskMetadata(
+        name="GLOBEV3Gender",
+        description="Gender classification from the GLOBE v3 dataset (sampled and enhanced from CommonVoice dataset for TTS purpose). This dataset is a stratified and downsampled version of the original dataset, containing about 535 hours of speech data across 164 accents. We use the gender column as the target label for audio classification.",
+        reference="https://huggingface.co/datasets/MushanW/GLOBE_V3",
+        dataset={
+            "path": "mteb/globe-v3-gender-mini",
+            "revision": "7020a6c14ec8a8e967013e04f2a695ead308bee1",
+        },
+        type="AudioClassification",
+        category="a2t",
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="accuracy",
+        date=("2025-05-26", "2025-05-26"),
+        domains=["Spoken", "Speech"],
+        task_subtypes=["Gender Classification"],
+        license="cc0-1.0",
+        annotations_creators="automatic",
+        dialect=[],
+        modalities=["audio"],
+        sample_creation="found",
+        bibtex_citation=r"""
+@misc{wang2024globe,
+  archiveprefix = {arXiv},
+  author = {Wenbin Wang and Yang Song and Sanjay Jha},
+  eprint = {2406.14875},
+  title = {GLOBE: A High-quality English Corpus with Global Accents for Zero-shot Speaker Adaptive Text-to-Speech},
+  year = {2024},
+}
+""",
+    )
+
+    input_column_name: str = "audio"
+    label_column_name: str = "predicted_gender"