paddlespeech.s2t.frontend.augmentor.noise_perturb module

Contains the noise perturb augmentation model.

class paddlespeech.s2t.frontend.augmentor.noise_perturb.NoisePerturbAugmentor(rng, min_snr_dB, max_snr_dB, noise_manifest_path)[source]

Bases: AugmentorBase

Augmentation model for adding background noise.

Parameters:

  • rng (random.Random) -- Random generator object.

  • min_snr_dB (float) -- Minimal signal noise ratio, in decibels.

  • max_snr_dB (float) -- Maximal signal noise ratio, in decibels.

  • noise_manifest_path (str) -- Manifest path for noise audio data.

Methods

__call__(x[, uttid, train])

Call self as a function.

transform_audio(audio_segment)

Add background noise audio.

transform_feature(spec_segment)

Adds various effects to the input audo feature segment.
transform_audio(audio_segment)[source]

Add background noise audio.

Note that this is an in-place transformation.

Parameters:

audio_segment (AudioSegmenet|SpeechSegment) -- Audio segment to add effects to.