Upload 7 files

.gitignore

@@ -0,0 +1,5 @@
+.venv/
+__pycache__/
+*.pyc
+.DS_Store
+Thumbs.db

CHANGELOG.md

@@ -0,0 +1,4 @@
+# Changelog
+
+## v1.0.0
+- Initial public scaffold: loader, scripts, benchmarks, docs (no data).

CITATION.cff

@@ -0,0 +1,10 @@
+cff-version: 1.2.0
+message: "If you use this dataset or code, please cite."
+title: "RIR-Mega"
+authors:
+  - family-names: Goswami
+    given-names: Mandip
+date-released: "2025-10-19"
+doi: "10.5281/zenodo.17387402"
+version: "1.0.0"
+repository-code: "https://github.com/mandip42/rirmega"

LICENSE

@@ -0,0 +1,2 @@
+CC BY 4.0 for metadata and docs; audio license is author-specified (e.g., CC BY-NC 4.0).
+

README.md

@@ -1,3 +1,48 @@
----
-license: cc-by-4.0
----
+# RIR-Mega
+
+RIR-Mega is a large-scale collection of room impulse responses (RIRs) with rich metadata for dereverberation, robust ASR, source localization, and room acoustics research.
+
+**This scaffold excludes data** (`data/audio/*`, `data/metadata/*.csv`/`.json`). Insert those locally before uploading to Hugging Face.
+
+- **Zenodo DOI (code/concept)**: https://doi.org/10.5281/zenodo.17387402
+- **GitHub (code)**: https://github.com/mandip42/rirmega
+
+## Layout
+```
+rirmega/
+├─ README.md
+├─ LICENSE
+├─ CHANGELOG.md
+├─ CITATION.cff
+├─ dataset_card.md
+├─ .gitattributes
+├─ .gitignore
+├─ rirmega/
+│  ├─ __init__.py
+│  ├─ dataset.py
+│  └─ schema.py
+├─ scripts/
+│  ├─ make_checksums.py
+│  ├─ validate_metadata.py
+│  ├─ to_sofa.py
+│  └─ upload_to_hf.py
+├─ benchmarks/
+│  ├─ rt60_regression/
+│  │  ├─ train_rt60.py
+│  │  └─ README.md
+│  └─ dereverb_sisdr/
+│     ├─ baseline_dereverb.py
+│     └─ README.md
+└─ data/
+   ├─ audio/        # (EXCLUDED in this zip — add your .wav/.flac)
+   └─ metadata/     # (EXCLUDED in this zip — add your metadata.csv/json)
+```
+
+## Quick start
+1. Add your `data/audio/*` and `data/metadata/metadata.csv` (and any folds/JSONs).
+2. (Optional) Run checksum + validation:
+   ```bash
+   python scripts/make_checksums.py
+   python scripts/validate_metadata.py data/metadata/metadata.csv
+   ```
+3. Upload to Hugging Face (choose Git or programmatic), see `scripts/upload_to_hf.py`.

SUBMITTING.md

@@ -0,0 +1,27 @@
+# Submitting to the RIR-Mega Leaderboard
+
+Thanks for evaluating on RIR-Mega! To submit a score, open a Pull Request that:
+1) Appends a row to the **Leaderboard** table in `dataset_card.md`.
+2) Includes the info below in the PR description (template provided).
+
+## Required details
+- **Method name + link** to code (GitHub/HF Space/Gist)
+- **Exact command** used, including `--target` if set
+- **Dataset tag** used (e.g., `v1.0.0`) and your commit hash if relevant
+- **Seed** (default baseline uses `random_state=0`)
+- **Train/Valid sizes** used (number of samples consumed)
+- **MAE (s)** and **RMSE (s)**
+
+## Reproducing the baseline
+```bash
+pip install soundfile numpy pandas scikit-learn
+python benchmarks/rt60_regression/train_rt60.py
+# or specify a target key that exists in metrics:
+python benchmarks/rt60_regression/train_rt60.py --target rt60
+```
+
+## Targets
+Common keys found in `metrics` (case-sensitive):  
+`rt60`, `drr_db`, `c50_db`, `c80_db`, `band_rt60s.125`, `band_rt60s.250`, `band_rt60s.500`, `band_rt60s.1000`, `band_rt60s.2000`, `band_rt60s.4000`
+
+If you use a different target, please document it clearly.

dataset_card.md

@@ -0,0 +1,81 @@
+# RIR-Mega
+
+Large-scale room impulse responses with rich metadata for dereverberation, robust ASR, localization, and room acoustics research.
+
+- **DOI**: https://doi.org/10.5281/zenodo.17387402  
+- **GitHub**: https://github.com/mandip42/rirmega
+
+## ✨ What’s inside
+- `data/` — RIR audio and `metadata/metadata.csv` (compact schema)
+- `rirmega/dataset.py` — Hugging Face Datasets loader
+- `benchmarks/rt60_regression/` — a lightweight RT60 regression baseline
+- `scripts/` — utilities (validation, checksums, mini subset)
+- *(optional)* `data-mini/` — tiny subset for quick demos and Spaces
+
+## 📦 Schema (compact)
+Required CSV columns:  
+`id, family, split, seed_room, fs, wav, room_size, absorption, absorption_bands, max_order, source, microphone, array, metrics, rng_seed`
+
+- `wav`: path to audio (relative to `data/` by default)
+- `fs`: sample rate (Hz)
+- `metrics`: JSON/dict-like string; may include keys like:  
+  `rt60`, `drr_db`, `c50_db`, `c80_db`, `band_rt60s.{125,250,500,1000,2000,4000}`
+
+## 🚀 Getting started
+```python
+from datasets import load_dataset
+ds = load_dataset("mandipgoswami/rirmega", name="default", trust_remote_code=True)
+print(ds)
+ex = ds["train"][0]
+audio = ex["audio"]  # dict with 'path' and array (on access)
+print(ex["sample_rate"], ex["file_path"])
+```
+
+## 🧪 Baseline: RT60 regression
+Lightweight features + RandomForest to predict RT60-like targets from RIR signals.
+
+```bash
+pip install soundfile numpy pandas scikit-learn
+python benchmarks/rt60_regression/train_rt60.py
+# or choose a specific target key present in `metrics`
+python benchmarks/rt60_regression/train_rt60.py --target rt60
+```
+**Default target search order:**  
+`rt60, drr_db, c50_db, c80_db, band_rt60s.125, 250, 500, 1000, 2000, 4000`
+
+### Reference numbers (example)
+- Train/Valid used: 36,000 / 4,000 (auto 10% valid)
+- Metric: MAE = **0.013 s**, RMSE = **0.022 s** (auto target)
+
+## 🏅 Leaderboard (RT60 regression)
+| Date | Team / Author | Method | Target | Train/Valid | MAE (s) | RMSE (s) | Seed | Code |
+|---|---|---|---|---|---:|---:|---:|---|
+| 2025-10-19 | Baseline (RIR-Mega) | RF on light feats | auto | 36k / 4k | 0.013 | 0.022 | 0 | `benchmarks/rt60_regression` |
+
+> 📫 **Submit a result:** Open a PR adding a row (see **Submitting**).
+
+## 📤 Submitting
+See **SUBMITTING.md** for rules and a PR template. Minimum info:
+- Command (incl. `--target` if used), seed, dataset tag (e.g., `v1.0.0`)
+- Train/Valid sizes used
+- MAE (s) and RMSE (s)
+- Link to code (repo, gist, or HF Space)
+
+## 🔖 Citation
+Please cite the dataset:
+```
+@dataset{RIRMega2025,
+  title={RIR-Mega},
+  author={Goswami, Mandip},
+  year={2025},
+  doi={10.5281/zenodo.17387402},
+  url={https://github.com/mandip42/rirmega}
+}
+```
+
+## 📜 Licenses
+- Metadata & docs: CC BY 4.0
+- Audio: specify here (e.g., CC BY-NC 4.0). If mixed, list per-subset.
+
+## 🏷️ Tags
+audio, rir, acoustics, dereverberation, robust-asr, simulation, room-acoustics