预处理训练集需要优化logmmse中的denoise函数
- Dominant language
- Python
- Stars
- 36.9k
- Forks
- 5.2k
- PR merge metrics
- No merged PRs in 30d
Description
**Summary[问题简述(一句话)]**
经过我的profile,我发现预处理训练集并不是因为IO慢,而是因为denoise拖慢了整个过程。我并不懂信号处理方面的知识,仅从我的专业知识来看,一是循环拖慢了速度,二是计算 广义指数积分(也就是denoise里用到的expn)的代价比较大。希望作者能优化denoise函数。
**Env & To Reproduce[复现与环境]**
描述你用的环境、代码版本、模型
我自己的仓库中https://github.com/0warning0error/MockingBird 修改了一点地方,为了找到拖慢预处理速度的原因做了profile。
**Screenshots[截图(如有)]**
If applicable, add screenshots to help

Contributor guide
No contributing guide indexed for this repository
Research direction
Locate the logmmse denoise function used during training-set preprocessing and inspect its loop and expn calculations. Profile that path to confirm the dominant cost, then verify that any optimization preserves denoising results while improving preprocessing time.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- performance
- Issue type
- Refactor
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100