paddlespeech.s2t.frontend.augmentor.volume_perturb module

Contains the volume perturb augmentation model.

class paddlespeech.s2t.frontend.augmentor.volume_perturb.VolumePerturbAugmentor(rng, min_gain_dBFS, max_gain_dBFS)[source]

Bases: AugmentorBase

Augmentation model for adding random volume perturbation.

This is used for multi-loudness training of PCEN. See

https://arxiv.org/pdf/1607.05666v1.pdf

for more details.

Parameters

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

  • min_gain_dBFS (float) -- Minimal gain in dBFS.

  • max_gain_dBFS (float) -- Maximal gain in dBFS.

Methods

__call__(x[, uttid, train])

Call self as a function.

transform_audio(audio_segment)

Change audio loadness.

transform_feature(spec_segment)

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

Change audio loadness.

Note that this is an in-place transformation.

Parameters

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