wav2vec2-xls-r-300m_phoneme-mfa_korean_nia13-asia-9634_001

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the NIA13 ASIA dataset.

Creator & Uploader: Jooyoung Lee (excalibur12@snu.ac.kr)

Training and evaluation data

Training Data

• Data Name: NIA13 ASIA
• Num. of Samples: 9,634
• Audio Length: 9H 42M

Test Data 1 (In-domain)

• Data Name: NIA13 ASIA
• Num. of Samples: 3,707
• Audio Length: 3H 37M

Test Data 2 (Out-of-domain)

• Data Name: SAMSUNG_60K
• Num. of Samples: 6,000
• Audio Length: 12 Hrs

Training hyperparameters

Training results

• Phone Error Rate on Test Data 1: 00.00%
• Phone Error Rate on Test Data 2: 00.00%

Framework versions

• Transformers 4.21.3
• Pytorch 1.12.1
• Datasets 2.4.0
• Tokenizers 0.12.1