Retrieval-based-Voice-Conversion-WebUI icon indicating copy to clipboard operation
Retrieval-based-Voice-Conversion-WebUI copied to clipboard

Off-by-one in extract_feature_print.py on macOS and numpy np.int

Open shaotz opened this issue 2 years ago • 0 comments

More information needed

Haven't tested on other platforms!

Changes:

  1. Using np.int64 instead of np.int in gui.py,extract_f0_print.py and vc_infer_pipeline.py to correspond to numpy's deprecating np.int.
  2. Added platform test in extract_feature_print.py , in that on macOS there's been one off-by-one error, causing version = sys.argv[6] to evaluate to out of range, i_gpu and exp_dir all getting the wrong value, and FileNotFoundError: [Errno 2] No such file or directory: '<PathTo>/3_feature256'(3_feature768) thereon.

-> 2) The problem hasn't been traced thoroughly, so I'm not sure what exactly the situation was. For my device, i_gpu and os.environ["CUDA_VISIBLE_DEVICES"] in the new branch could be omitted(commented out), leaving it the same as in if len(sys.argv) == 5

    exp_dir = sys.argv[4]
    version = sys.argv[5]

off-by-one

shaotz avatar May 20 '23 15:05 shaotz