Implementation:Speechbrain Speechbrain Prepare IEMOCAP

Knowledge Sources	SpeechBrain
Domains	Emotion_Recognition, Data_Preparation
Last Updated	2026-02-09 00:00 GMT

Overview

Concrete tool for preparing the IEMOCAP dataset for emotion recognition provided by the SpeechBrain library.

Description

This script creates JSON data manifest files for the IEMOCAP (Interactive Emotional Dyadic Motion Capture) dataset. It parses session directories, extracts emotion labels from evaluation files, segments audio utterances, and supports two splitting strategies: random ratio-based splits and speaker-disjoint splits for leave-two-speaker-out cross-validation across 10 speakers. The dataset contains 5,531 utterances with categorical emotion annotations.

Usage

Use this when preparing the IEMOCAP dataset for emotion recognition training with SpeechBrain recipes.

Code Reference

Source Location

• Repository: SpeechBrain
• File: recipes/IEMOCAP/emotion_recognition/iemocap_prepare.py

Signature

def prepare_data(
    data_original,
    save_json_train,
    save_json_valid,
    save_json_test,
    split_ratio=[80, 10, 10],
    different_speakers=False,
    test_spk_id=1,
    seed=12,
):

Import

from recipes.IEMOCAP.emotion_recognition.iemocap_prepare import prepare_data

I/O Contract

Inputs

Name	Type	Required	Description
data_original	str	Yes	Path to the folder where the original IEMOCAP dataset is stored
save_json_train	str	Yes	Path where the train data specification file will be saved
save_json_valid	str	Yes	Path where the validation data specification file will be saved
save_json_test	str	Yes	Path where the test data specification file will be saved
split_ratio	list	No	Ratios for train/valid/test splits (default: [80, 10, 10])
different_speakers	bool	No	If True, ensure speakers are not shared among splits (default: False)
test_spk_id	int	No	Speaker ID for test set in leave-two-speaker-out mode (default: 1)
seed	int	No	Random seed for reproducibility (default: 12)

Outputs

Name	Type	Description
train.json	JSON	Training split manifest with audio paths and emotion labels
valid.json	JSON	Validation split manifest
test.json	JSON	Test split manifest

Usage Examples

from recipes.IEMOCAP.emotion_recognition.iemocap_prepare import prepare_data

prepare_data(
    data_original="/path/to/IEMOCAP",
    save_json_train="train.json",
    save_json_valid="valid.json",
    save_json_test="test.json",
    split_ratio=[80, 10, 10],
    different_speakers=False,
)

Related Pages

• Principle:Speechbrain_Speechbrain_Dataset_Specific_Data_Preparation