---
pretty_name: NADI 2026 SLU Test Set
configs:
- config_name: default
  data_files:
  - split: clean
    path: data/clean-*
dataset_info:
  features:
  - name: audio
    dtype: audio
  - name: id
    dtype: string
  splits:
  - name: clean
    num_examples: 989
---

# NADI 2026 Official Test Set

Official test set for NADI 2026 Subtask 5: Spoken Language Understanding.

Repository:

https://huggingface.co/datasets/UBC-NLP/nadi26_subtask_5_SLU_test

The dataset contains the `clean` test split with 989 examples.

Each example includes:

- `audio`
- `id`

The audio is embedded directly in Parquet.