speechbrain.lobes.models

Package defining neural netword models (CRDNN, Xvectors …)

speechbrain.lobes.models.CRDNN

A combination of Convolutional, Recurrent, and Fully-connected networks.

speechbrain.lobes.models.ContextNet

The SpeechBrain implementation of ContextNet by https://arxiv.org/pdf/2005.03191.pdf

speechbrain.lobes.models.ECAPA_TDNN

A popular speaker recognition and diarization model.

speechbrain.lobes.models.ESPnetVGG

This lobes replicate the encoder first introduced in ESPNET v1

speechbrain.lobes.models.MetricGAN

Generator and discriminator used in MetricGAN

speechbrain.lobes.models.MetricGAN_U

Generator and discriminator used in MetricGAN-U

speechbrain.lobes.models.RNNLM

Implementation of a Recurrent Language Model.

speechbrain.lobes.models.VanillaNN

Vanilla Neural Network for simple tests.

speechbrain.lobes.models.Xvector

A popular speaker recognition and diarization model.

speechbrain.lobes.models.conv_tasnet

Implementation of a popular speech separation model.

speechbrain.lobes.models.convolution

This is a module to ensemble a convolution (depthwise) encoder with or without residule connection.

speechbrain.lobes.models.dual_path

Library to support dual-path speech separation.

speechbrain.lobes.models.fairseq_wav2vec

This lobe enables the integration of fairseq pretrained wav2vec models.

speechbrain.lobes.models.huggingface_wav2vec

This lobe enables the integration of huggingface pretrained wav2vec2 models.

speechbrain.lobes.models.segan_model

This file contains two PyTorch modules which together consist of the SEGAN model architecture (based on the paper: Pascual et al. https://arxiv.org/pdf/1703.09452.pdf).

speechbrain.lobes.models.transformer

High level processing blocks.