Usage: ds_onnx_infer [-h] --ds-file VAR --acoustic-config VAR --vocoder-config VAR
       [--spk VAR] [--transpose VAR] [--dur] [--variance] [--pitch] --out VAR
       [--depth VAR] [--steps VAR]
       [--ep VAR] [--device-index VAR] [--verbose]

Optional arguments:
  -h, --help            shows help message and exits
  -v, --version         prints version information and exits
  --ds-file             Path to .ds file [required]
  --acoustic-config     Path to acoustic dsconfig.yaml [required]
  --vocoder-config      Path to vocoder.yaml [required]
  --spk                 Speaker Mixture (e.g. "name" or "name1|name2" or "name1:0.25|name2:0.75")
  --transpose           Transpose pitch in semitones [default: 0]
  --dur                 Use Duration Predictor
  --variance            Use Multi-Variance Predictor
  --pitch               Use Pitch Predictor
  --out                 Output Audio Filename (*.wav) [required]
  --depth               Shallow diffusion depth (type: int, range: [1, 1000])
                        or Rectified flow depth (type: float, range: [0, 1]) [default: 1]
  --ep                  Execution Provider for audio inference. (cpu/directml/cuda)
                        [default: "cpu"]
  --device-index        GPU device index [default: 0]
  --verbose             Show debug-level logs