microsoft/VibeVoice

Largest clones in the code

• 45 lines in 2 placespython396 tokens
  vibevoice/processor/vibevoice_processor.py:646–690·vibevoice/processor/vibevoice_streaming_processor.py:363–407

  646def batch_decode(self, *args, **kwargs):
  647    """
  648    This method forwards all its arguments to VibeVoiceTextTokenizer's [`~PreTrainedTokenizer.batch_decode`].
  649    Please refer to the docstring of this method for more information.
  650    """
  651    return self.tokenizer.batch_decode(*args, **kwargs)
  652
  653def decode(self, *args, **kwargs):
  654    """
  655    This method forwards all its arguments to VibeVoiceTextTokenizer's [`~PreTrainedTokenizer.decode`].
  

  First 10 of 45 lines.

• 51 lines in 2 placespython368 tokens
  vibevoice/processor/vibevoice_processor.py:461–511·vibevoice/processor/vibevoice_streaming_processor.py:313–363

  461def prepare_speech_inputs(
  462    self,
  463    speech_inputs: List[np.ndarray],
  464    return_tensors: Optional[Union[str, TensorType]] = None,
  465    device: Optional[Union[str, torch.device]] = None,
  466    dtype: Optional[torch.dtype] = None,
  467) -> Dict[str, Any]:
  468    """
  469    Prepare speech inputs for model consumption.
  470
  

  First 10 of 51 lines.

• 78 lines in 2 placespython365 tokens
  vibevoice/processor/vibevoice_processor.py:55–132·vibevoice/processor/vibevoice_streaming_processor.py:51–128

  55    [`VibeVoiceProcessor`]: The processor object instantiated from pretrained model.
  56"""
  57import os
  58import json
  59from transformers.utils import cached_file
  60from .vibevoice_tokenizer_processor import VibeVoiceTokenizerProcessor
  61from vibevoice.modular.modular_vibevoice_text_tokenizer import (
  62    VibeVoiceTextTokenizer,
  63    VibeVoiceTextTokenizerFast
  64)
  

  First 10 of 78 lines.