Implementation:Speechbrain Speechbrain Train UrbanSound8k

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

Overview

Concrete tool for training sound class embeddings on the UrbanSound8k dataset provided by the SpeechBrain library.

Description

This recipe defines the UrbanSound8kBrain class (subclass of sb.core.Brain) for training sound class embedding models (x-vectors or ECAPA-TDNN) on the UrbanSound8k dataset. The pipeline extracts spectral features, optionally applies amplitude-to-dB conversion and mean-variance normalization, then feeds the features through an embedding model and classifier. Supports waveform augmentation with label replication for augmented samples. Evaluation includes per-fold accuracy tracking and confusion matrix generation.

Usage

Use this recipe to train a sound classifier on the UrbanSound8k dataset using either x-vector or ECAPA-TDNN architectures. Requires the UrbanSound8k data folder. Configure with hparams/train_ecapa_tdnn.yaml or hparams/train.yaml.

Code Reference

Source Location

• Repository: SpeechBrain
• File: recipes/UrbanSound8k/SoundClassification/train.py

Signature

class UrbanSound8kBrain(sb.core.Brain):
    def compute_forward(self, batch, stage):
        ...
    def compute_objectives(self, predictions, batch, stage):
        ...

Import

python recipes/UrbanSound8k/SoundClassification/train.py hparams/train_ecapa_tdnn.yaml --data_folder /path/to/UrbanSound8K

I/O Contract

Inputs

Name	Type	Required	Description
batch	PaddedBatch	Yes	Batch containing sig (waveforms) and class_string_encoded (class labels)
stage	sb.Stage	Yes	TRAIN, VALID, or TEST

Outputs

Name	Type	Description
predictions	tuple	Classifier output logits and lens
loss	torch.Tensor	NLL classification loss

Usage Examples

python train.py hparams/train_ecapa_tdnn.yaml --data_folder /path/to/UrbanSound8K

Related Pages

• Principle:Speechbrain_Speechbrain_Sound_Classification_Training