Spaces:
Running
on
Zero
GPU task aborted
help me ;w;
@RuneLightLovely Hello! Thank you for using our project. Please pay close attention to the duration of the audio you input:
In Fast Mode, the audio should not exceed 4 seconds to avoid timing out. Audio inputs longer than 4 seconds will be automatically trimmed to 4 seconds.
In Quality Mode, the GPU duration is dynamically calculated as: 60s (preprocessing) + video_seconds × denoising_steps × 3 seconds. For example, an 8-second audio with 25 denoising steps will take approximately 660 seconds (60 + 8 × 25 × 3 = 60 + 600 = 660s ≈ 11 minutes). You can adjust the denoising steps in advanced options to accommodate longer audio durations or reduce GPU time.
If you wish to process longer videos, feel free to visit https://github.com/HKUST-C4G/AnyTalker for unlimited duration generation!
O- ok, i get it, for now...b- but, other people might not know this either, not just me...but why are you telling me this now, and not before!?
(note: this is my HONEST opinion, so I hope you understand... :/ and also, I hope you don't get mad at me ;w;)
O- ok, i get it, for now...b- but, other people might not know this either, not just me...but why are you telling me this now, and not before!?
(note: this is my HONEST opinion, so I hope you understand... :/ and also, I hope you don't get mad at me ;w;)
We are constantly working on improving the project, and your feedback is incredibly valuable to us. The demo is more robust now, and we truly appreciate your contribution. Thank you for being honest and understanding :)
We've now added prominent reminders in the demo and announcements, and have set the program to allocate reasoning time dynamically. This way, future users can avoid these issues.