Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
A wav2vec2.0-xlsr-53 model is fine-tuned on a large speech corpus of children aged 3 to 5 years.
|
| 2 |
+
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
---
|
| 6 |
+
language:
|
| 7 |
+
- ko
|
| 8 |
+
thumbnail: "url to a thumbnail used in social sharing"
|
| 9 |
+
tags:
|
| 10 |
+
- automatic-speech-recognition
|
| 11 |
+
- speech
|
| 12 |
+
datasets:
|
| 13 |
+
- AI-HUB
|
| 14 |
+
metrics:
|
| 15 |
+
- cer
|
| 16 |
+
model-index:
|
| 17 |
+
- name: xlsr-53-korean_kids_3_to_5_syl
|
| 18 |
+
results:
|
| 19 |
+
- task:
|
| 20 |
+
type: automatic-speech-recognition
|
| 21 |
+
dataset:
|
| 22 |
+
type: AI-HUB
|
| 23 |
+
name: AI-HUB
|
| 24 |
+
split: test
|
| 25 |
+
metrics:
|
| 26 |
+
- type: cer
|
| 27 |
+
value: 3.48
|
| 28 |
+
name: Syllable Error Rate
|
| 29 |
+
---
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
Dataset: children's speech aged 3 to 5 years
|
| 33 |
+
- the number of files) train:valid:test=256,548:12,121:12,000
|
| 34 |
+
- transcripts were tokenized into morphs by Mecab
|
| 35 |
+
|
| 36 |
+
EVALUATION results
|
| 37 |
+
- Dataset: testset from the corpus
|
| 38 |
+
- Performance: syllable Error Rate on the testset = 3.48%
|
| 39 |
+
|