nemo_curator.stages.audio.preprocessing.mono_conversion

View as Markdown

Audio mono conversion stage.

Converts multi-channel audio to mono and verifies sample rate. Typically the first stage in an audio processing pipeline.

Module Contents

Classes

NameDescription
MonoConversionStageAudio mono conversion and sample rate verification stage.

API

class nemo_curator.stages.audio.preprocessing.mono_conversion.MonoConversionStage(
output_sample_rate: int = 48000,
audio_filepath_key: str = 'audio_filepath',
strict_sample_rate: bool = True,
name: str = 'MonoConversion',
batch_size: int = 1,
resources: nemo_curator.stages.resources.Resources = (lambda: Resources(cpus=1.0...
)
Dataclass

Bases: ProcessingStage[AudioTask, AudioTask]

Audio mono conversion and sample rate verification stage.

Converts multi-channel audio to mono by averaging channels. Optionally verifies that audio matches expected sample rate.

Parameters:

output_sample_rate
intDefaults to 48000

Expected sample rate in Hz (default: 48000)

audio_filepath_key
strDefaults to 'audio_filepath'

Key in data dict for audio file path

strict_sample_rate
boolDefaults to True

If True, reject audio with wrong sample rate

audio_filepath_key
str = 'audio_filepath'
batch_size
int = 1
name
str = 'MonoConversion'
output_sample_rate
int = 48000
resources
Resources
strict_sample_rate
bool = True
nemo_curator.stages.audio.preprocessing.mono_conversion.MonoConversionStage.__post_init__()
nemo_curator.stages.audio.preprocessing.mono_conversion.MonoConversionStage.inputs() -> tuple[list[str], list[str]]
nemo_curator.stages.audio.preprocessing.mono_conversion.MonoConversionStage.outputs() -> tuple[list[str], list[str]]
nemo_curator.stages.audio.preprocessing.mono_conversion.MonoConversionStage.process(
task: nemo_curator.tasks.AudioTask
) -> nemo_curator.tasks.AudioTask | list[nemo_curator.tasks.AudioTask]

Convert audio to mono and verify sample rate.

Mutates task.data in-place with waveform data. Returns task if successful, [] if doesn’t meet requirements.