modelscope / FunASR

A Fundamental End-to-End Speech Recognition Toolkit and Open Source SOTA Pretrained Models, Supporting Speech Recognition, Voice Activity Detection, Text Post-processing etc.
https://www.funasr.com
Other
6.46k stars 687 forks source link

请问paraformer训练时对数据有时长限制吗,我下了一个数据集,是20多个音频文件,时长都是几十分钟一个,训练之后测试得到的模型效果和训练前一样 #1896

Closed Yoki185 closed 3 months ago

Yoki185 commented 3 months ago

Notice: In order to resolve issues more efficiently, please raise issue following the template. (注意:为了更加高效率解决您遇到的问题,请按照模板提问,补充细节)

❓ Questions and Help

请问paraformer训练时对数据有时长限制吗,我使用的数据集,是20多个音频文件,时长都是几十分钟一个,训练之后测试得到的模型效果和训练前一样

Before asking:

  1. search the issues.
  2. search the docs.

What is your question?

Code

What have you tried?

What's your environment?

RERUIN76 commented 1 month ago

这个问题您似乎已经解决了,可以分享一下解决方法吗,我现在微调时也是训练后和训练前推理效果相同,ps:我只训练了五条5秒左右的数据。是否会是因为数据过少的原因。但是训练后的模型去推理训练时喂给的数据都不能正确识别。