Commit b6e5af78 authored by Phil Wang's avatar Phil Wang
Browse files

optionally allow for resampling directly within SoundDataset, if target_sample_khz specified

parent f7756f56

audiolm_pytorch/data.py

+7 −1
Original line number Diff line number Diff line
@@ -21,6 +21,7 @@ class SoundDataset(Dataset): 
        folder,

        exts = ['flac', 'wav'],

        max_length = None,

        target_sample_khz = None,

        seq_len_multiple_of = None

    ):

        super().__init__()
@@ -32,6 +33,8 @@ class SoundDataset(Dataset): 


        self.files = files

        self.max_length = max_length



        self.target_sample_khz = target_sample_khz

        self.seq_len_multiple_of = seq_len_multiple_of



    def __len__(self):
@@ -39,10 +42,13 @@ class SoundDataset(Dataset): 


    def __getitem__(self, idx):

        file = self.files[idx]

        data, samplerate = torchaudio.load(file)

        data, sample_khz = torchaudio.load(file)



        data = rearrange(data, '1 ... -> ...')



        if exists(self.target_sample_khz):

            data = torchaudio.functional.resample(data, sample_khz, self.target_sample_khz)



        if exists(self.max_length):

            data = data[:self.max_length]

setup.py

+1 −1
Original line number Diff line number Diff line
@@ -3,7 +3,7 @@ from setuptools import setup, find_packages 
setup(

  name = 'audiolm-pytorch',

  packages = find_packages(exclude=[]),

  version = '0.0.23',

  version = '0.0.24',

  license='MIT',

  description = 'AudioLM - Language Modeling Approach to Audio Generation from Google Research - Pytorch',

  author = 'Phil Wang',