Checklist ONIA 2026 Team Selection Camp (Extended Squad) – Test 2 · Task 3
The Talkative Robots
Romanian title: Roboții vorbăreți
Group 300 synthetic speech recordings into triplets of the same message and identify which of three speech-synthesis systems produced each test recording.
The task
Three robotic factions (IronClad, Prism, Genesis) each use a distinct speech-synthesis algorithm, possibly with several voice profiles per faction. In task 1, 300 WAV files contain 100 messages, each read once by each of the three robots; the contestant must group the files into 100 triplets with identical spoken words.
In task 2, 900 labelled training recordings (300 per faction) are provided, and the contestant must assign each of 400 test recordings to its faction (1 = IronClad, 2 = Prism, 3 = Genesis).
Abridged and translated by SOTA from the official Romanian materials. The official statement has the exact rules, and it wins wherever this summary differs.
At a glance
- You get
- task1/ (300 .wav files); task2/train/ with
task2_train.csv(semicolon-separated name;class) and task2/test/ (400 .wav files). Archive of 759,651,687 bytes. - You submit
- CSV with datapointID, subtaskID and answer: 100 rows for subtask 1 (answer = three file names joined by '|') and 400 rows for subtask 2 (datapointID = test file name, answer = 1, 2 or 3).
- Scoring
- Task 1 (40 points): 40 × correct triplets / 100 (invalid triplets score 0). Task 2 (60 points): 60 × accuracy over 400 files; accuracy ≥ 0.85 receives full marks.
- Rules
- On-site, Timișoara (West University of Timișoara and Politehnica University of Timișoara)
- No internet access; only preinstalled offline resources (NLTK, spaCy
en_core_web_sm, scikit-learn, Gensim glove-wiki-gigaword-50, fastText without pretrained models, and the Hugging Face models distilbert-base-uncased, MoritzLaurer DeBERTa-v3-large-mnli-fever-anli-ling-wanli, roberta-large-mnli, all-MiniLM-L6-v2 and whisper-small) - A GPU Jupyter environment hosted by UVT was provided during the tests
- Problem statements unlocked with a password handed out in the room
- Format
- ONIA 2026 Lot Lărgit, test 2, 24 May 2026.