hhoangphuoc commited on
Commit
09b4186
·
verified ·
1 Parent(s): be1af83

Update dataset descriptions

Browse files
Files changed (1) hide show
  1. README.md +41 -0
README.md CHANGED
@@ -30,4 +30,45 @@ configs:
30
  path: data/validation-*
31
  - split: test
32
  path: data/test-*
 
 
 
 
 
 
 
33
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30
  path: data/validation-*
31
  - split: test
32
  path: data/test-*
33
+ task_categories:
34
+ - automatic-speech-recognition
35
+ tags:
36
+ - paralinguistic
37
+ pretty_name: a
38
+ size_categories:
39
+ - 100K<n<1M
40
  ---
41
+ A preprocessed version of `Switchboard Corpus`. The corpus audio has been upsampled to 16kHz, separated channels and the transcripts have been processed
42
+ with special treats for paralinguistic events, particularly laughter and speech-laughs.
43
+ This preprocessed dataset has been processed for ASR task. For the original dataset, please check out the original link: https://catalog.ldc.upenn.edu/LDC97S62
44
+
45
+ The dataset has been splitted into train, test and validation sets with 70/20/10 ratio, as following summary:
46
+
47
+ ```python
48
+ Train Dataset (70%): Dataset({
49
+ features: ['audio', 'sampling_rate', 'transcript'],
50
+ num_rows: 185402
51
+ })
52
+ Validation Dataset (10%): Dataset({
53
+ features: ['audio', 'sampling_rate', 'transcript'],
54
+ num_rows: 20601
55
+ })
56
+ Test Dataset (20%): Dataset({
57
+ features: ['audio', 'sampling_rate', 'transcript'],
58
+ num_rows: 51501
59
+ })
60
+ ```
61
+
62
+ An example of the content is this dataset:
63
+ ```
64
+ ```
65
+
66
+ Regarding the total amount of laughter and speech-laugh existing in the dataset, here is the overview:
67
+ ```bash
68
+ Train Dataset (swb_train): {'laughter': 16044, 'speechlaugh': 9586}
69
+
70
+ Validation Dataset (swb_val): {'laughter': 1845, 'speechlaugh': 1133}
71
+
72
+ Test Dataset (swb_test): {'laughter': 4335, 'speechlaugh': 2775}
73
+ ```
74
+