Checklist AICC 2025 Round 1 · Task 3
Is that audio?
Predict an anonymised language label for short speech clips drawn from heterogeneous public speech datasets.
The task
The clips were assembled from public speech datasets, and each carries an anonymised class label (remapped to dataset-specific anonymised IDs). Quality, loudness, background noise, speaker accent and recording conditions vary between clips.
The training metadata list clip paths and their target language labels, which may include noise or mislabels. Sampling rates, durations and volume levels vary. Every test clip belongs to one of the classes present in the training set.
Contestants predict the label of each test clip using only the provided training data.
Abridged by SOTA from the official materials. The official statement has the exact rules, and it wins wherever this summary differs.
At a glance
- You get
train.csv(path, label),test.csv(ID, path) and the referenced .wav files.- You submit
- CSV with columns ID and label.
- Scoring
- Accuracy (Kaggle metric 'Accuracy Score').
- Rules
- Only the provided training data may be used.
- Individual participation (maximum team size 1); at most 20 submissions per day.
- AICC contest rules (stated on each Kaggle rules page, not enforceable): no use of LLMs for writing code or getting task ideas; no internet use other than official library documentation and the contest platform; no communication with anyone during the contest; clarifications only via the #clarification-requests channel on the AICC Discord server.
- Format
- AICC Round 1 (November 2025), online on Kaggle; Kaggle competition window 15 Nov 2025 (about 18:50 UTC) – 16 Nov 2025 19:00 UTC.