Dancing2Music icon indicating copy to clipboard operation
Dancing2Music copied to clipboard

Output video length not match the input audio length

Open HsiaoYingLu opened this issue 5 years ago • 9 comments

Hi, After modifying the output feature number to 28 in mfcc_rev.py, which is not the correct way to get the feature according to the paper, I can successfully execute demo.py. I run demo.py and use a 4:02 .wav audio file as input. But the program outputs 72096 frames, which makes the output video length to be 1:20:xx under fps=15 setting. Why is the length of output video much longer than the length of input audio?

Any thought on this will be helpful. Thanks!

HsiaoYingLu avatar Jul 16 '20 02:07 HsiaoYingLu

@HsiaoYingLu hi, would you mind sharing aud_3cls.ckpt I do not find it './data/stats/aud_3cls.ckpt', or the author said that The ckpt is under the downloaded "stats" folder.

thanks

Ai-is-light avatar Jul 22 '20 04:07 Ai-is-light

You may follow the steps below to get the folder "stats" and "aud_3cls.ckpt" within.

URL=http://vllab.ucmerced.edu/hylee/Dancing2Music/data.zip wget -N $URL -O ./data.zip unzip ./data.zip -d . rm ./data.zip These instructions are provided by the author @HsinYingLee

HsiaoYingLu avatar Jul 22 '20 07:07 HsiaoYingLu

@HsiaoYingLu Thanks, I didn't download the data.zip

Ai-is-light avatar Jul 22 '20 15:07 Ai-is-light

I put a zipped file of the complete demo code on my personal webpage.

HsinYingLee avatar Jul 30 '20 03:07 HsinYingLee

@HsiaoYingLu were you able to solve this issue, also got the same problem

ak9250 avatar Sep 18 '20 04:09 ak9250

Hi, After modifying the output feature number to 28 in mfcc_rev.py, which is not the correct way to get the feature according to the paper, I can successfully execute demo.py. I run demo.py and use a 4:02 .wav audio file as input. But the program outputs 72096 frames, which makes the output video length to be 1:20:xx under fps=15 setting. Why is the length of output video much longer than the length of input audio?

Any thought on this will be helpful. Thanks!

Hi, I was wondering if you could solve this issue?I am facing a same problem

Nooshin-github avatar Apr 29 '21 12:04 Nooshin-github

Hi, After modifying the output feature number to 28 in mfcc_rev.py, which is not the correct way to get the feature according to the paper, I can successfully execute demo.py. I run demo.py and use a 4:02 .wav audio file as input. But the program outputs 72096 frames, which makes the output video length to be 1:20:xx under fps=15 setting. Why is the length of output video much longer than the length of input audio?

Any thought on this will be helpful. Thanks!

hi i am facing the same problem. could you tell me how to modify the mfcc_rev.py? can you run the the demo.zip provided by @HsinYingLee successfully? i met lots of bug when reproducing this demo..... thank you very much.

learningyan avatar May 31 '21 12:05 learningyan

Hi, After modifying the output feature number to 28 in mfcc_rev.py, which is not the correct way to get the feature according to the paper, I can successfully execute demo.py. I run demo.py and use a 4:02 .wav audio file as input. But the program outputs 72096 frames, which makes the output video length to be 1:20:xx under fps=15 setting. Why is the length of output video much longer than the length of input audio? Any thought on this will be helpful. Thanks!

Hi, I was wondering if you could solve this issue?I am facing a same problem

hi have you solved this problem?

learningyan avatar May 31 '21 12:05 learningyan

Hi try this

pip3 install ffmpegpip3 install ffmpeg-normalizepython demo.py --decomp_snapshot Stage1.ckpt --comp_snapshot Stage2.ckpt --aud_path ChillingMusic.wav --out_file demo/out4.mp4 --out_dir demo/out_frame4

if the video length is 3 times more than the audio length go to the demo.py file and change the line : total_t = int(length/32+1) to this total_t = int(length/32/3+1) I hope it helps.

On Mon, May 31, 2021 at 2:35 PM QixinYan @.***> wrote:

Hi, After modifying the output feature number to 28 in mfcc_rev.py, which is not the correct way to get the feature according to the paper, I can successfully execute demo.py. I run demo.py and use a 4:02 .wav audio file as input. But the program outputs 72096 frames, which makes the output video length to be 1:20:xx under fps=15 setting. Why is the length of output video much longer than the length of input audio? Any thought on this will be helpful. Thanks!

Hi, I was wondering if you could solve this issue?I am facing a same problem

hi have you solved this problem?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/NVlabs/Dancing2Music/issues/9#issuecomment-851460782, or unsubscribe https://github.com/notifications/unsubscribe-auth/ARPLNXNSVFGZLHQ3WEIJDZDTQN67JANCNFSM4O3L6QGQ .

Nooshin-github avatar May 31 '21 12:05 Nooshin-github