V33DA: A Multi-Modal Benchmark for Vocal Attribution in Freely Interacting Zebra Finches

Task. Given a detected zebra finch vocalization and the set of birds visible at that moment, determine which bird produced the call.

V33DA provides 33,625 vocalization events from 10 individually identified zebra finches across 3 experiments (2021--2023), each with synchronized 5-channel audio, multi-view video, 3D pose, FM radio telemetry, and accelerometer-derived ground-truth labels.

	BP01	BP02	BP05	Total
Birds	4	3	3	10
Events	16,219	11,577	5,829	33,625
Recording days	10	3	3	16
Raw hours	~46h	~14h	~14h	~74h

Code: github.com/marisbasha/v33da

Quick start

pip install huggingface_hub
huggingface-cli download songbirdini/V33DA --repo-type dataset --local-dir data/v33da

Or in Python:

from huggingface_hub import snapshot_download
snapshot_download("songbirdini/V33DA", repo_type="dataset", local_dir="data/v33da")

Loading a sample

Each binary array column is serialized with numpy.save. To decode:

import io
import numpy as np
import pyarrow.parquet as pq

table = pq.read_table("data/v33da/v33da-00000.parquet")
row = table.to_pydict()

# 3D keypoints: (N_birds, 5, 3) float64
kp3d = np.load(io.BytesIO(row["keypoints_3d"][0]))

# Audio path -> multichannel WAV
audio_path = row["audio_path"][0]  # e.g. "audio/juvExpBP01/2021-06-28/..."

The included explore.ipynb notebook shows how to inspect audio, video, pose, radio, and accelerometer data, and how to reproject 3D keypoints into 2D camera views using the shipped calibrations.

Modalities

Each parquet row is one vocalization event:

Column	Description
audio_path	5-channel cage-microphone WAV (24,414 Hz, float32)
accelerometer_path	Per-bird accelerometer WAV (oracle verification signal)
video_path	Aligned composite MP4 clip (3 views, 47.68 fps)
keypoints_3d	Triangulated 3D pose: 5 keypoints per bird (beak, head, backpack, tail base, tail tip)
keypoints_2d_top	2D keypoints in top-view image space
keypoints_2d_back	2D keypoints in back-view image space
radio_*	21 per-bird FM radio telemetry signals per frame
vocalizer_idx	Ground-truth caller index
bird_color	Color identity of the vocalizer
experiment	Experiment name (juvExpBP01 / juvExpBP02 / juvExpBP05)
date	Recording date

Ground-truth labels

Labels come from on-body accelerometer vibration, not from microphone-based localization. A WhisperSeg model proposes candidate windows from the demodulated signal; events are retained only when audible in microphones, showing characteristic on-body vibration, with no overlapping call from another bird. All candidates are manually reviewed.

Filtering

From 39,329 reviewed candidates:

• 1,768 rejected as invalid during manual review
• 3,784 removed by overlap filtering (+/-10 ms)
• 152 removed by 3D-to-2D reprojection error (> 40 px)

Yielding 33,625 released events.

Per-bird counts

Experiment	Bird	Events
BP01	blue	5,735
BP01	peach	2,277
BP01	red	3,963
BP01	white	4,244
BP02	blue	4,923
BP02	peach	4,589
BP02	red	2,065
BP05	brown	3,633
BP05	purple	2,140
BP05	yellow	56

Release contents

Path	Description
v33da-*.parquet	23 pooled parquet shards
audio/	Multichannel microphone WAV files
accelerometer/	Multichannel accelerometer WAV files
clips/	Aligned composite MP4 clips
calibrations/	Per-experiment camera calibration files
explore.ipynb	Dataset exploration notebook
metadata.json	Release schema and filter metadata

Evaluation regimes

The benchmark code supports three evaluation regimes:

Regime	What transfers?	Use case
Session-disjoint	Same birds, different days	In-domain ceiling
Held-out experiment	New birds, new year, new cage	Cross-cohort transfer
Leave-one-bird-out	One unseen bird per fold	Identity-shift stress test

Limitations

V33DA is collected in one aviary with one recording geometry across three experiments. Cross-experiment evaluation combines bird-identity shift, group-composition shift, and recording-date shift. The released benchmark excludes overlapping vocalizations and filters out samples with poor geometric consistency.