Too Slowly

[wangchunself]

This is a bit slow to produce results.

It took over 30 minutes to run the audio for over 1 minute, and it still didn't work. My 4090 graphics card...

[Honee-W]

Thank you for your feedback!

This inference speed is abnormally slow. While long audio usually uses segmentation + stitching, even unsegmented inference shouldn’t take this long.

To narrow down the issue first, could you check:

1. Is GPU acceleration enabled? (Verify with torch.cuda.is_available() for PyTorch.)
2. What inference batch size are you using? (Too small may waste GPU capacity.)

Once you have the code, share error logs, nvidia-smi GPU usage, or profiling results if possible—this will speed up diagnosis.

Sorry for the inconvenience, and thanks for your patience!

[wangchunself]

I'm not sure if you have the habit of checking emails. I sent an email to your email address. Please take a look.

[wangchunself]

this email: zq_wang@mail.nwpu.edu.cn