diart.functional#

Module Contents#

Functions#

overlapped_speech_penalty(segmentation[, gamma, beta])

normalize_embeddings(embeddings[, norm])

diart.functional.overlapped_speech_penalty(segmentation, gamma=3, beta=10)#
Parameters:

  • segmentation (torch.Tensor) –

  • gamma (float) –

  • beta (float) –

diart.functional.normalize_embeddings(embeddings, norm=1)#
Parameters:

  • embeddings (torch.Tensor) –

  • norm (float | torch.Tensor) –

Return type:

torch.Tensor