search

sstzal / DFRF

[ECCV2022] The implementation for "Learning Dynamic Facial Radiance Fields for Few-Shot Talking Head Synthesis".
MIT License
335 stars 43 forks source link

poor performance on lip sync for chinese video #12

Open yxqyyy opened 1 year ago

yxqyyy commented 1 year ago

Thanks for your work! I want to train it with Chinese video,however, the output video having a poor performance on lip sync, is this because the deepspeech model does not support Chinese? Or is there something wrong with my training?

sstzal commented 1 year ago

Thanks for your work! I want to train it with Chinese video,however, the output video having a poor performance on lip sync, is this because the deepspeech model does not support Chinese? Or is there something wrong with my training?

deepspeech支持中文，但是它训练集中的中文数据确实比较少，所以可能会对结果有一些影响，但是应该不会太差。

yxqyyy commented 1 year ago

Thanks for your work! I want to train it with Chinese video,however, the output video having a poor performance on lip sync, is this because the deepspeech model does not support Chinese? Or is there something wrong with my training?

deepspeech支持中文，但是它训练集中的中文数据确实比较少，所以可能会对结果有一些影响，但是应该不会太差。

感谢解答，可否提供一个你们预训练好的中文模型啊？

essen900718 commented 8 months ago

@yxqyyy 哈囉想請問你最後有訓練出比較好的結果嗎?