Implementation:Speechbrain Speechbrain Train CommonLanguage LangId

Knowledge Sources	SpeechBrain
Domains	ASR, Training
Last Updated	2026-02-09 00:00 GMT

Overview

Concrete tool for training a language identification system on the CommonLanguage dataset provided by the SpeechBrain library.

Description

This recipe defines the LID class (subclass of sb.Brain) for spoken language identification. The pipeline includes a prepare_features method that handles waveform augmentation, feature extraction, and normalization. The compute_forward method passes features through an embedding model (e.g., ECAPA-TDNN) and a classifier to produce language class probabilities. NLL loss is used for training, and error rate statistics are tracked during evaluation.

Usage

Use this recipe to train a language identification model using the ECAPA-TDNN architecture on the CommonLanguage dataset. Requires the corresponding hyperparameter YAML file and data preparation script.

Code Reference

Source Location

• Repository: SpeechBrain
• File: recipes/CommonLanguage/lang_id/train.py

Signature

class LID(sb.Brain):
    def prepare_features(self, wavs, stage):
        """Prepare the features for computation, including augmentation."""
        ...
    def compute_forward(self, batch, stage):
        """Runs all the computation of that transforms the input into the
        output probabilities over the N classes."""
        ...
    def compute_objectives(self, inputs, batch, stage):
        """Computes the loss given the predicted and targeted outputs."""
        ...

Import

# Run as recipe script
python recipes/CommonLanguage/lang_id/train.py hparams/train_ecapa_tdnn.yaml --data_folder /path/to/CommonLanguage

I/O Contract

Inputs

Name	Type	Required	Description
batch.sig	torch.Tensor	Yes	Input waveform signal
batch.language_encoded	torch.Tensor	Yes	Encoded language label

Outputs

Name	Type	Description
outputs	torch.Tensor	Class posterior probabilities over N language classes
lens	torch.Tensor	Relative signal lengths

Usage Examples

python train.py hparams/train_ecapa_tdnn.yaml --data_folder /path/to/CommonLanguage

Related Pages

• Principle:Speechbrain_Speechbrain_Sound_Classification_Training