Datasets:
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -15,8 +15,19 @@ tags:
|
|
| 15 |
size_categories:
|
| 16 |
- 10K<n<100K
|
| 17 |
configs:
|
| 18 |
-
- config_name:
|
| 19 |
default: true
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
data_files:
|
| 21 |
- split: train
|
| 22 |
path: "NonSpeech7k/train/*.tar"
|
|
@@ -50,7 +61,10 @@ A collection of human non-speech vocal sound datasets in WebDataset format, usef
|
|
| 50 |
|
| 51 |
## Subsets
|
| 52 |
|
| 53 |
-
###
|
|
|
|
|
|
|
|
|
|
| 54 |
- **Samples**: 7,014 (train: 6,289, test: 725)
|
| 55 |
- **Classes**: Breathing, Coughing, Crying, Laughing, Screaming, Sneezing, Yawning
|
| 56 |
- **Source**: [Zenodo](https://zenodo.org/records/6967442)
|
|
@@ -102,13 +116,12 @@ All datasets are stored in WebDataset format:
|
|
| 102 |
```python
|
| 103 |
from datasets import load_dataset
|
| 104 |
|
| 105 |
-
# Load
|
| 106 |
ds = load_dataset("gijs/speech-utterances")
|
| 107 |
|
| 108 |
# Load a specific subset
|
| 109 |
ds = load_dataset("gijs/speech-utterances", "Expresso")
|
| 110 |
-
ds = load_dataset("gijs/speech-utterances", "
|
| 111 |
-
ds = load_dataset("gijs/speech-utterances", "EmoGator")
|
| 112 |
|
| 113 |
# Access individual examples
|
| 114 |
print(ds['train'][0])
|
|
|
|
| 15 |
size_categories:
|
| 16 |
- 10K<n<100K
|
| 17 |
configs:
|
| 18 |
+
- config_name: all
|
| 19 |
default: true
|
| 20 |
+
data_files:
|
| 21 |
+
- split: train
|
| 22 |
+
path:
|
| 23 |
+
- "NonSpeech7k/train/*.tar"
|
| 24 |
+
- "NonSpeech7k/test/*.tar"
|
| 25 |
+
- "VocalSound/*.tar"
|
| 26 |
+
- "DeeplyNonverbalVocalization/*.tar"
|
| 27 |
+
- "EmoGator/*.tar"
|
| 28 |
+
- "Expresso/*.tar"
|
| 29 |
+
- "VIVAE/*.tar"
|
| 30 |
+
- config_name: NonSpeech7k
|
| 31 |
data_files:
|
| 32 |
- split: train
|
| 33 |
path: "NonSpeech7k/train/*.tar"
|
|
|
|
| 61 |
|
| 62 |
## Subsets
|
| 63 |
|
| 64 |
+
### all (default)
|
| 65 |
+
All datasets concatenated together (~75k samples total).
|
| 66 |
+
|
| 67 |
+
### NonSpeech7k
|
| 68 |
- **Samples**: 7,014 (train: 6,289, test: 725)
|
| 69 |
- **Classes**: Breathing, Coughing, Crying, Laughing, Screaming, Sneezing, Yawning
|
| 70 |
- **Source**: [Zenodo](https://zenodo.org/records/6967442)
|
|
|
|
| 116 |
```python
|
| 117 |
from datasets import load_dataset
|
| 118 |
|
| 119 |
+
# Load all datasets concatenated (default)
|
| 120 |
ds = load_dataset("gijs/speech-utterances")
|
| 121 |
|
| 122 |
# Load a specific subset
|
| 123 |
ds = load_dataset("gijs/speech-utterances", "Expresso")
|
| 124 |
+
ds = load_dataset("gijs/speech-utterances", "NonSpeech7k")
|
|
|
|
| 125 |
|
| 126 |
# Access individual examples
|
| 127 |
print(ds['train'][0])
|