---
language:
- ch
license: apache-2.0
base_model: openai/whisper-small
datasets:
- TingChen-ppmc/Shanghai_Dialect_Conversational_Speech_Corpus
metrics:
- wer
- cer
---
This is a finetuned version of Whisper-small on Shanghai_Dialect_Conversational_Speech_Corpus
Results
CER: 56.07
WER: 75.62
Chinese Word Error Rate (WER) is based on Jieba Word segmentation