Upload metadata.json.txt with huggingface_hub

#1
by imadhavan - opened
Files changed (1) hide show
  1. metadata.json.txt +43 -0
metadata.json.txt ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "OpenSER-700K",
3
+ "description": "OpenSER-700K is a large-scale, high-quality speech emotion recognition dataset containing 700,258 audio samples across 7 emotion categories (Angry, Disgust, Fear, Happy, Neutral, Sad, Surprise). It is curated for deep learning, multimodal fusion, and benchmarking in affective computing and human-computer interaction research.",
4
+ "version": "1.0.0",
5
+ "total_audio_files": 700258,
6
+ "duration": "≈1200 hours (avg. 6 seconds per clip)",
7
+ "sampling_rate": 16000,
8
+ "audio_format": "wav",
9
+ "classes": {
10
+ "angry": 98477,
11
+ "disgust": 98477,
12
+ "fear": 98477,
13
+ "happy": 98477,
14
+ "neutral": 98505,
15
+ "sad": 98477,
16
+ "surprise": 99348
17
+ },
18
+ "modalities": ["audio", "speech"],
19
+ "recommended_usage": [
20
+ "Training speech emotion recognition (SER) models",
21
+ "Feature extraction with HuBERT, Wav2Vec2, or OpenL3",
22
+ "Multimodal fusion with facial emotion embeddings",
23
+ "Benchmarking speech-based affect recognition tasks"
24
+ ],
25
+ "dataset_structure": [
26
+ "OpenSER-700K/{emotion}/{filename}.wav",
27
+ "Metadata file: data.json (maps file path → emotion label)"
28
+ ],
29
+ "license": "cc-by-nc-4.0",
30
+ "usage_notes": "This dataset is for academic and non-commercial research only. Please provide attribution when using it in publications or derivative works.",
31
+ "source_datasets": [
32
+ {"name": "RAVDESS", "license": "CC BY-NC"},
33
+ {"name": "SAVEE", "license": "Research-use-only"},
34
+ {"name": "TESS", "license": "Research-use-only"},
35
+ {"name": "CREMA-D", "license": "CC BY-NC"},
36
+ {"name": "EMO-DB", "license": "Research-use-only"},
37
+ {"name": "IEMOCAP", "license": "Research-use-only"}
38
+ ],
39
+ "libraries": ["datasets", "torchaudio", "transformers"],
40
+ "formats": ["wav"],
41
+ "tags": ["speech-emotion-recognition", "audio", "SER", "emotion", "deep-learning"],
42
+ "citation": "Adhavan M. OpenSER-700K: Large-Scale Speech Emotion Recognition Dataset. 2025. URL: https://huggingface.co/datasets/imadhavan/OpenSER-700K"
43
+ }