espnet2.gan_codec.shared.loss.freq_loss.MultiScaleMelSpectrogramLoss

source

class espnet2.gan_codec.shared.loss.freq_loss.MultiScaleMelSpectrogramLoss(fs: int = 22050, range_start: int = 6, range_end: int = 11, window: str = 'hann', n_mels: int = 80, fmin: int | None = 0, fmax: int | None = None, center: bool = True, normalized: bool = False, onesided: bool = True, log_base: float | None = 10.0, alphas: bool = True)

Bases: Module

Multi-Scale spectrogram loss.

• Parameters:
  • fs (int) – Sampling rate.
  • range_start (int) – Power of 2 to use for the first scale.
  • range_stop (int) – Power of 2 to use for the last scale.
  • window (str) – Window type.
  • n_mels (int) – Number of mel bins.
  • fmin (Optional *[*int ]) – Minimum frequency for Mel.
  • fmax (Optional *[*int ]) – Maximum frequency for Mel.
  • center (bool) – Whether to use center window.
  • normalized (bool) – Whether to use normalized one.
  • onesided (bool) – Whether to use oneseded one.
  • log_base (Optional *[*float ]) – Log base value.
  • alphas (bool) – Whether to use alphas as coefficients or not..

Initializes internal Module state, shared by both nn.Module and ScriptModule.

forward(y_hat: Tensor, y: Tensor) → Tensor

Calculate Mel-spectrogram loss.

• Parameters:
  • y_hat (Tensor) – Generated waveform tensor (B, 1, T).
  • y (Tensor) – Groundtruth waveform tensor (B, 1, T).
• Returns: Mel-spectrogram loss value.
• Return type: Tensor