Datasets:
Camera-ready card: authors, radio definitions, unevaluated companion, NeurIPS citation
Browse files
README.md
CHANGED
|
@@ -35,6 +35,10 @@ configs:
|
|
| 35 |
|
| 36 |
# V33DA++: Overlapping-Call and Long-Context Companion to V33DA
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
V33DA++ is the companion release to the [V33DA](https://huggingface.co/datasets/songbirdini/V33DA) benchmark. It is built from the same BirdPark recordings and the same 10 zebra finches, but covers data **outside** the strict single-vocalizer attribution clips that define V33DA. V33DA++ is intended for tasks that benefit from **overlapping callers** or **longer temporal context**: source separation, vocal activity detection, audio-visual synchronization, and active-speaker detection.
|
| 39 |
|
| 40 |
V33DA++ ships **three buckets**, each as a separate config:
|
|
@@ -45,7 +49,7 @@ V33DA++ ships **three buckets**, each as a separate config:
|
|
| 45 |
| `padded` | 33,625 | ±2 s context windows around V33DA single-vocalizer events. Original event onsets and offsets are preserved inside the longer window, supplying long-range context without altering the underlying call. |
|
| 46 |
| `overlap_padded` | 3,784 | The `overlap` bucket with the same ±2 s context, for long-context multi-caller tasks. |
|
| 47 |
|
| 48 |
-
> V33DA++ is **
|
| 49 |
|
| 50 |
**Code:** [github.com/marisbasha/v33da](https://github.com/marisbasha/v33da) · **Main benchmark:** [songbirdini/V33DA](https://huggingface.co/datasets/songbirdini/V33DA)
|
| 51 |
|
|
@@ -148,7 +152,7 @@ Each parquet row is one vocalization event. Most binary array columns are NumPy
|
|
| 148 |
|---|---|---|
|
| 149 |
| `audio` | bytes | NumPy-serialized `(5, T)` float32 array — 5 cage microphones at 24,414 Hz, padded by `context_pad_sec` on each side |
|
| 150 |
| `audio_path` | string | Path to a separately exported multichannel WAV mirror (best-effort, see Release contents) |
|
| 151 |
-
| `accelerometer` | bytes | NumPy-serialized `(N_birds, T)` float32 array — per-bird on-body vibration channel at 24,414 Hz, used
|
| 152 |
| `accelerometer_path` | string | Path to a separately exported multichannel WAV mirror (best-effort) |
|
| 153 |
| `video_path` | string | Path to a composite multi-view MP4 (top, back, side) — `null` if not available |
|
| 154 |
|
|
@@ -166,16 +170,19 @@ Per-frame indexing: bird `i`'s keypoint `k` at frame `t` is `kp3d[t, i, k]` (and
|
|
| 166 |
|
| 167 |
### Radio telemetry (every visible bird, every frame)
|
| 168 |
|
| 169 |
-
Per-bird
|
| 170 |
|
| 171 |
| Field | Description |
|
| 172 |
|---|---|
|
| 173 |
-
| `radio_frq` |
|
| 174 |
-
| `radio_thetaS` |
|
| 175 |
-
| `radio_thetaA / B / C / D` | Per-antenna
|
| 176 |
-
| `radio_phiA / B / C / D
|
| 177 |
-
| `
|
| 178 |
-
| `
|
|
|
|
|
|
|
|
|
|
| 179 |
|
| 180 |
### Overlap-event metadata (`overlap` and `overlap_padded` only)
|
| 181 |
|
|
@@ -233,10 +240,10 @@ CC-BY-4.0. External tools used during preprocessing (e.g., Whisper / WhisperSeg,
|
|
| 233 |
If you use V33DA or V33DA++, please cite the V33DA paper:
|
| 234 |
|
| 235 |
```bibtex
|
| 236 |
-
@
|
| 237 |
-
title
|
| 238 |
-
author
|
| 239 |
-
|
| 240 |
-
|
| 241 |
}
|
| 242 |
```
|
|
|
|
| 35 |
|
| 36 |
# V33DA++: Overlapping-Call and Long-Context Companion to V33DA
|
| 37 |
|
| 38 |
+
**Authors:** Maris Basha¹, Yuhang Wang¹, Xiaoran Chen², Longbiao Cheng¹, Luca Yapura¹, Anja T. Zai¹, Mathieu Salzmann², Richard Hahnloser¹
|
| 39 |
+
¹ Institute of Neuroinformatics, University of Zurich and ETH Zurich · ² Swiss Data Science Center and CVLab, EPFL
|
| 40 |
+
**Maintainer:** Maris Basha · **Provenance:** same BirdPark recordings, pipeline, and animal license (ZH148/2020) as V33DA.
|
| 41 |
+
|
| 42 |
V33DA++ is the companion release to the [V33DA](https://huggingface.co/datasets/songbirdini/V33DA) benchmark. It is built from the same BirdPark recordings and the same 10 zebra finches, but covers data **outside** the strict single-vocalizer attribution clips that define V33DA. V33DA++ is intended for tasks that benefit from **overlapping callers** or **longer temporal context**: source separation, vocal activity detection, audio-visual synchronization, and active-speaker detection.
|
| 43 |
|
| 44 |
V33DA++ ships **three buckets**, each as a separate config:
|
|
|
|
| 49 |
| `padded` | 33,625 | ±2 s context windows around V33DA single-vocalizer events. Original event onsets and offsets are preserved inside the longer window, supplying long-range context without altering the underlying call. |
|
| 50 |
| `overlap_padded` | 3,784 | The `overlap` bucket with the same ±2 s context, for long-context multi-caller tasks. |
|
| 51 |
|
| 52 |
+
> V33DA++ is an **unevaluated companion release**: no result in the V33DA paper depends on it. For the single-vocalizer attribution benchmark, use [V33DA](https://huggingface.co/datasets/songbirdini/V33DA).
|
| 53 |
|
| 54 |
**Code:** [github.com/marisbasha/v33da](https://github.com/marisbasha/v33da) · **Main benchmark:** [songbirdini/V33DA](https://huggingface.co/datasets/songbirdini/V33DA)
|
| 55 |
|
|
|
|
| 152 |
|---|---|---|
|
| 153 |
| `audio` | bytes | NumPy-serialized `(5, T)` float32 array — 5 cage microphones at 24,414 Hz, padded by `context_pad_sec` on each side |
|
| 154 |
| `audio_path` | string | Path to a separately exported multichannel WAV mirror (best-effort, see Release contents) |
|
| 155 |
+
| `accelerometer` | bytes | NumPy-serialized `(N_birds, T)` float32 array — per-bird on-body vibration channel at 24,414 Hz, used to verify the caller; withheld from benchmark inputs |
|
| 156 |
| `accelerometer_path` | string | Path to a separately exported multichannel WAV mirror (best-effort) |
|
| 157 |
| `video_path` | string | Path to a composite multi-view MP4 (top, back, side) — `null` if not available |
|
| 158 |
|
|
|
|
| 170 |
|
| 171 |
### Radio telemetry (every visible bird, every frame)
|
| 172 |
|
| 173 |
+
Per-bird BirdRadio signals from a 4-antenna receiver (antennas A–D; M = master vector). Each is a NumPy `(N_birds, T_radio)` float32 byte blob with one value per radio frame (512 samples at 24,414 Hz = one video frame). Symbols follow the BirdPark paper (Rüttimann et al., PeerJ 2025); see the [V33DA card](https://huggingface.co/datasets/songbirdini/v33da#radio-telemetry).
|
| 174 |
|
| 175 |
| Field | Description |
|
| 176 |
|---|---|
|
| 177 |
+
| `radio_frq` | PLL tracking frequency |
|
| 178 |
+
| `radio_thetaS` | Master-vector phase angle $\theta_i$ (deg) |
|
| 179 |
+
| `radio_thetaA / B / C / D` | Per-antenna rotated-vector angles $\alpha_i^a$ (deg) |
|
| 180 |
+
| `radio_phiA / B / C / D` | Per-antenna compensation phases $\phi_i^a$ (deg) |
|
| 181 |
+
| `radio_phiM` | Sum of the compensation phases (deg) |
|
| 182 |
+
| `radio_powRA / B / C / D / M` | Baseband signal power (dBFS) |
|
| 183 |
+
| `radio_powNA / B / C / D / M` | Baseband signal variance (dBFS) |
|
| 184 |
+
|
| 185 |
+
BirdRadio does not compute an angle of arrival.
|
| 186 |
|
| 187 |
### Overlap-event metadata (`overlap` and `overlap_padded` only)
|
| 188 |
|
|
|
|
| 240 |
If you use V33DA or V33DA++, please cite the V33DA paper:
|
| 241 |
|
| 242 |
```bibtex
|
| 243 |
+
@inproceedings{basha2026v33da,
|
| 244 |
+
title = {Who Called? V33DA: A Physically Verified Multimodal Benchmark for Vocal Attribution in Zebra Finch Groups},
|
| 245 |
+
author = {Basha, Maris and Wang, Yuhang and Chen, Xiaoran and Cheng, Longbiao and Yapura, Luca and Zai, Anja T. and Salzmann, Mathieu and Hahnloser, Richard},
|
| 246 |
+
booktitle = {Advances in Neural Information Processing Systems (NeurIPS), Evaluations and Datasets Track},
|
| 247 |
+
year = {2026},
|
| 248 |
}
|
| 249 |
```
|