Datasets:

Modalities:
Audio
Text
Formats:
parquet
Languages:
Croatian
Size:
< 1K
Libraries:
Datasets
pandas
5roop commited on
Commit
8fe879e
1 Parent(s): edf16f4

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +30 -3
README.md CHANGED
@@ -13,13 +13,13 @@ dataset_info:
13
  dtype: string
14
  splits:
15
  - name: test
16
- num_bytes: 7993646.0
17
  num_examples: 95
18
  - name: train
19
- num_bytes: 68421127.0
20
  num_examples: 277
21
  download_size: 76394317
22
- dataset_size: 76414773.0
23
  configs:
24
  - config_name: default
25
  data_files:
@@ -27,4 +27,31 @@ configs:
27
  path: data/test-*
28
  - split: train
29
  path: data/train-*
 
 
 
 
 
30
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
  dtype: string
14
  splits:
15
  - name: test
16
+ num_bytes: 7993646
17
  num_examples: 95
18
  - name: train
19
+ num_bytes: 68421127
20
  num_examples: 277
21
  download_size: 76394317
22
+ dataset_size: 76414773
23
  configs:
24
  - config_name: default
25
  data_files:
 
27
  path: data/test-*
28
  - split: train
29
  path: data/train-*
30
+ language:
31
+ - hr
32
+ pretty_name: Mići Princ
33
+ size_categories:
34
+ - n<1K
35
  ---
36
+
37
+ # Mići Princ - Chakavian ASR Dataset
38
+
39
+ Antoine de Saint-Exupéry's _The Little Prince_, translated into Chakavian dialect of Croatian and read by native speakers.
40
+
41
+ This version is split into chunks of [0, 30] seconds for ASR experiments.
42
+
43
+ ## Dataset composition
44
+
45
+ Attributes:
46
+ * audio: audio chunk. In `train`, the audio can include multiple speakers and has a duration in [0, 30] seconds for easier ASR training. In `test`, only one speaker is present in an instance.
47
+ * text: Text with punctuation, newlines, bullets, and special characters.
48
+ * normalized_text: Text with some punctuation removed and special characters remapped for easier training. Characters, not present in standard Croatian (ȅ,é,ȋ,ȉ,ȃ,ȁ), are substituted with non-special characters (e,i,a). Bullets, newlines, and quotation marks are removed.
49
+ * speaker: In `train`, this attribute is always an empty string. In `test` it accurately represents the speaker of the chunk.
50
+
51
+ ## Authors:
52
+
53
+ * Nikola Ljubešić [https://huggingface.co/nljubesi](https://huggingface.co/nljubesi)
54
+ * Peter Rupnik [https://huggingface.co/5roop][https://huggingface.co/5roop]
55
+ * Tea Perinčić
56
+
57
+