modules/voice_conversion/fairseq/data/offset_tokens_dataset.py

# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.

from . import BaseWrapperDataset


class OffsetTokensDataset(BaseWrapperDataset):
    def __init__(self, dataset, offset):
        super().__init__(dataset)
        self.offset = offset

    def __getitem__(self, idx):
        return self.dataset[idx] + self.offset