search

fschmid56 / cpjku_dcase23

This repository contains the code of the CP JKU submission to DCASE23 Task 1 "Low-complexity Acoustic Scene Classification"
22 stars 4 forks source link

cpjku_dcase23 #1

Closed Stirve587 closed 10 months ago

Stirve587 commented 1 year ago

cpjku_dcase23 Could you tell me how to solve this problem?

fschmid56 commented 1 year ago

It seems that you don't have the file _fold1train.csv in the correct location. Make sure you correctly download and unzip all the files from this URL: https://zenodo.org/record/6337421

The missing file _fold1train.csv is specifically located in this archive: https://zenodo.org/record/6337421/files/TAU-urban-acoustic-scenes-2022-mobile-development.meta.zip?download=1

Stirve587 commented 1 year ago

I'm sorry to bother you again. I'm a newbie and I hope you can bear with me. How can I solve this mistake 1MK~B32WDZI2EQMNQJ8XK0K

fschmid56 commented 1 year ago

Hi, no problem, and thanks for your interest.

In general, resampling the waveform is time-consuming and we only want to do it once - therefore we cache the resampled waveforms. The command line argument _cachepath specifies in which directory the resampled waveform ends up. Per default _cachepath was set to datasets/cpath which caused problems on your OS. I changed it to os.path.join("datasets", "cpath") . I hope this fixes your problem.

Stirve587 commented 1 year ago

After I run the run_training.py file,I still get these errors, 5D09A3C{2$)PBLCCWB({682 BZ}_Q0~TRG5O}J$UXC2 KYL Z3$)JC$F}H(~PHX 8U(617A YU9N$@T 1EG1 F_8M XK%AC I hope you can look at it for me when you have time. Thank you very much

fschmid56 commented 1 year ago

Have you already resolved it? For sure it has something to do with building the paths on your systems (the "\" and "/" are not matching).

Stirve587 commented 1 year ago

I have solved the previous problem, and there are still some mistakes after the correction QQ图片20231019152213 QQ图片20231019152251

fschmid56 commented 1 year ago

It seems like the expression len(self.dirs) is 0. Can you verify that and check why this is the case?

Stirve587 commented 1 year ago

For your original code, I have changed this part according to your readme, and have also modified these two parts, but I have not modified other codes of yours. May I ask why I have encountered this problem and how can I modify it? QQ图片20231022142809 Thank you very much for your reply QQ图片20231022142841

Stirve587 commented 1 year ago

I am very sorry that data set 4 was not put in due to my negligence, which caused some errors,but now there is a new problem, I have checked some information but still have not solved, I hope you can give me some advice, thank you very QQ图片20231025142229 much

Stirve587 commented 1 year ago

My dear author, do you have time to reply me? I feel that I am about to get rid of this problem. I hope you can reply as soon as possible

Stirve587 commented 1 year ago

Thanks for your help these days, I have been running successfully

fschmid56 commented 1 year ago

Hi! I'm glad to hear that the code works for you now. I apologize for the lack of responsiveness, I have been very busy lately.

Stirve587 commented 1 year ago

You are the best author I have ever seen. I hope you can achieve greater achievements in your field. Thank you for your guidance

Stirve587 commented 1 year ago

ZY4HYZ8 WROB%WQ_O~_X)03 I have such a problem after running, how can I solve it

fschmid56 commented 1 year ago

I haven't encountered this specific error so far. Is the training correctly logged to wandb? Or is any information lost?

Augenstern1120 commented 10 months ago

Have you ever encountered this kind of problem：RuntimeError: stack expects each tensor to be equal size, but got [1, 32000] at entry 0 and [1, 320000] at entry 46

fschmid56 commented 10 months ago

I'm afraid, I have not encountered this problem before. Can you provide more information on where the problem occurs? It seems that you are dealing with a 1-second and a 10-second audio file.