nemo_curator.stages.audio.tagging.resample_audio

Resample Audio Stage

Resamples audio files to a target sample rate and format. Follows the exact pattern from NeMo Curator: https://github.com/NVIDIA-NeMo/Curator/blob/main/nemo_curator/stages/audio/common.py

Module Contents

Classes

Name	Description
`ResampleAudioStage`	Stage for resampling audio files in a TTS/ALM dataset.

API

class nemo_curator.stages.audio.tagging.resample_audio.ResampleAudioStage(
    resampled_audio_dir: str,
    input_format: str = 'wav',
    target_sample_rate: int = 16000,
    target_format: str = 'wav',
    target_nchannels: int = 1,
    audio_filepath_key: str = 'audio_filepath',
    resampled_audio_filepath_key: str = 'resampled_audio_filepath',
    duration_key: str = 'duration',
    audio_item_id_key: str = 'audio_item_id',
    name: str = 'ResampleAudio'
)

Dataclass

Bases: ProcessingStage[AudioTask, AudioTask]

Stage for resampling audio files in a TTS/ALM dataset.

Takes a manifest containing audio file paths and resamples them to target sample rate and format, while creating a new manifest with updated paths.

audio_filepath_key

str = 'audio_filepath'

audio_item_id_key

str = 'audio_item_id'

duration_key

str = 'duration'

input_format

str = 'wav'

name

str = 'ResampleAudio'

resampled_audio_dir

str

resampled_audio_filepath_key

str = 'resampled_audio_filepath'

target_format

str = 'wav'

target_nchannels

int = 1

target_sample_rate

int = 16000

nemo_curator.stages.audio.tagging.resample_audio.ResampleAudioStage.inputs() -> tuple[list[str], list[str]]

nemo_curator.stages.audio.tagging.resample_audio.ResampleAudioStage.outputs() -> tuple[list[str], list[str]]

nemo_curator.stages.audio.tagging.resample_audio.ResampleAudioStage.process(
    task: nemo_curator.tasks.AudioTask
) -> nemo_curator.tasks.AudioTask

Process a single task by resampling the audio file.

Parameters:

task

AudioTask

AudioTask with data dict containing audio_filepath and audio_item_id(optional)

Returns: AudioTask

AudioTask with updated metadata

nemo_curator.stages.audio.tagging.resample_audio.ResampleAudioStage.setup_on_node(
    _node_info: nemo_curator.backends.base.NodeInfo | None = None,
    _worker_metadata: nemo_curator.backends.base.WorkerMetadata | None = None
) -> None