Add ACIDS public catalog (10 models, ~1.6 GB)

Mirrors the canonical anonymous-download set from
https://play.forum.ircam.fr/rave-vst-api/get_available_models:
VCTK, darbouka_onnx, nasa, percussion, vintage, isis, musicnet,
sol_ordinario, sol_full, sol_ordinario_fast.

Each .ts ships a matching .json sidecar (same schema as the IIL set).
README rewritten to cover both the IIL-curated mirror and this new set.
All CC-BY-NC-4.0 inherited from upstream ACIDS.

.gitattributes

@@ -51,3 +51,13 @@ voice-multi-b2048-r48000-z11.ts filter=lfs diff=lfs merge=lfs -text
 voice_hifitts_b2048_r48000_z16.ts filter=lfs diff=lfs merge=lfs -text
 voice_jvs_b2048_r44100_z16.ts filter=lfs diff=lfs merge=lfs -text
 voice_vctk_b2048_r44100_z22.ts filter=lfs diff=lfs merge=lfs -text
+VCTK.ts filter=lfs diff=lfs merge=lfs -text
+darbouka_onnx.ts filter=lfs diff=lfs merge=lfs -text
+isis.ts filter=lfs diff=lfs merge=lfs -text
+musicnet.ts filter=lfs diff=lfs merge=lfs -text
+nasa.ts filter=lfs diff=lfs merge=lfs -text
+percussion.ts filter=lfs diff=lfs merge=lfs -text
+sol_full.ts filter=lfs diff=lfs merge=lfs -text
+sol_ordinario.ts filter=lfs diff=lfs merge=lfs -text
+sol_ordinario_fast.ts filter=lfs diff=lfs merge=lfs -text
+vintage.ts filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,28 +1,131 @@
+---
+license: cc-by-nc-4.0
+tags:
+- audio
+- rave
+- timbre-transfer
+- neural-synthesis
+- ircam
+- maestro
+language:
+- en
+pipeline_tag: audio-to-audio
+---
+
 # RAVE — AEmotionStudio mirror
 
-Curated mirror of [Intelligent-Instruments-Lab/rave-models](https://huggingface.co/Intelligent-Instruments-Lab/rave-models) RAVE checkpoints, used by MAESTRO's RAVE Timbre Transfer panel (opt-in starter pack).
+Curated mirror of public **RAVE** (Realtime Audio Variational autoEncoder) checkpoints, used
+by MAESTRO's RAVE Timbre Transfer panel (opt-in starter pack). Sources:
+
+- The [Intelligent-Instruments-Lab/rave-models](https://huggingface.co/Intelligent-Instruments-Lab/rave-models) curated set (birds, voices, organs, water, etc.).
+- The [official ACIDS-IRCAM public catalog](https://acids-ircam.github.io/rave_models_download.html), pulled from the canonical anonymous API at `https://play.forum.ircam.fr/rave-vst-api/get_available_models`.
+
+RAVE was developed by [Antoine Caillon](https://caillonantoine.github.io/) and the
+[ACIDS team at IRCAM](https://www.ircam.fr/). Paper: [arXiv:2111.05011](https://arxiv.org/abs/2111.05011).
+Upstream code: [acids-ircam/RAVE](https://github.com/acids-ircam/RAVE).
+
+## License
+
+**CC-BY-NC-4.0** — non-commercial use only, inherited from the upstream distributions.
+Generated audio is fine for non-commercial use. Commercial use of the *models themselves*
+(e.g. shipping them inside a paid product) requires permission from the original authors / IRCAM.
+
+Per MAESTRO's stance (see `LICENSE_AUDIT.md` and the `feedback_download_on_demand_licensing`
+memory), these weights are fetched *on demand* by the end user — the user (not MAESTRO the
+binary) is the licensee.
 
-License: **CC-BY-NC-4.0** — non-commercial use only. See the upstream model card for full terms.
+---
 
-## Files
+## Models — IIL-curated set (b2048 streaming exports, 18 models)
 
+Each `.ts` checkpoint has a `<stem>.json` sidecar with name, license, sample-rate, latent-dim,
+source URL, and a one-line description.
+
+### Voice / speech
 - `voice_vocalset_b2048_r48000_z16.ts` — **Voice (VocalSet)**. Voice timbre trained on the VocalSet corpus — covers vocal techniques across multiple singers. Use for the canonical 'make this sound like a voice' transfer.
 - `voice-multi-b2048-r48000-z11.ts` — **Voice (Multi-speaker)**. Aggregated multi-speaker voice corpus. Wider speaker diversity than VocalSet — produces more 'average human' renders.
-- `voice_hifitts_b2048_r48000_z16.ts` — **Voice (HiFi-TTS)**. HiFi-TTS — high-fidelity expressive English speech corpus. Cleaner, more articulate than the multi-speaker model.
-- `voice_jvs_b2048_r44100_z16.ts` — **Voice (JVS, Japanese)**. JVS — Japanese multi-speaker voice corpus at 44.1 kHz. Use for Japanese-language sources or non-Latin phoneme structure.
-- `voice_vctk_b2048_r44100_z22.ts` — **Voice (VCTK, English)**. VCTK — English multi-speaker corpus from CSTR Edinburgh, 44.1 kHz. High 22-dim latent — captures more speaker idiosyncrasies.
+- `voice_hifitts_b2048_r48000_z16.ts` — **Voice (HiFi-TTS)**. High-fidelity expressive English speech corpus. Cleaner, more articulate than the multi-speaker model.
+- `voice_jvs_b2048_r44100_z16.ts` — **Voice (JVS, Japanese)**. JVS Japanese multi-speaker corpus at 44.1 kHz. Use for Japanese-language sources or non-Latin phoneme structure.
+- `voice_vctk_b2048_r44100_z22.ts` — **Voice (VCTK, English)**. VCTK English multi-speaker corpus from CSTR Edinburgh, 44.1 kHz. High 22-dim latent — captures more speaker idiosyncrasies.
+
+### Bird / wildlife
 - `birds_motherbird_b2048_r48000_z16.ts` — **Birds (Motherbird)**. Bird-vocalization corpus — chirps + textural transients. The canonical 'weird' pick: produces wildly warped output for any arbitrary input.
 - `birds_dawnchorus_b2048_r48000_z8.ts` — **Birds (Dawn Chorus)**. Dense overlapping bird vocalizations recorded at dawn. Smaller 8-dim latent — outputs lean ensemble-textural over individual calls.
 - `birds_pluma_b2048_r48000_z12.ts` — **Birds (Pluma)**. Lighter, individual bird-call timbres. Mid-size 12-dim latent balances character + clarity.
 - `humpbacks_pondbrain_b2048_r48000_z20.ts` — **Humpback Whales**. Humpback-whale song. Long, slow, hauntingly-deep vocal contours — pairs well with sustained input.
 - `marinemammals_pondbrain_b2048_r48000_z20.ts` — **Marine Mammals**. Mixed marine-mammal vocalizations — dolphins, orcas, sea-life clicks and cries.
+
+### Instruments
 - `guitar_iil_b2048_r48000_z16.ts` — **Guitar (IIL)**. Acoustic / electric guitar timbre. Good demo for transferring voice or synth input into a plucked-string voice.
 - `organ_bach_b2048_r48000_z16.ts` — **Organ (Bach)**. Pipe-organ timbre trained on Bach repertoire. Sustained harmonic textures — pairs well with melodic input.
 - `organ_archive_b2048_r48000_z16.ts` — **Organ (Archive)**. Historical pipe-organ recordings — broader, dustier textures than the Bach model. Good for film-score atmospheres.
 - `sax_soprano_franziskaschroeder_b2048_r48000_z20.ts` — **Soprano Sax (Schroeder)**. Soprano-saxophone extended techniques by Franziska Schroeder. Multiphonics, growls, key clicks. 20-dim latent — captures fine-grained articulation.
+- `mrp_strengjavera_b2048_r44100_z16.ts` — **Magnetic Resonator Piano (Strengjavera)**. Sustained metallic-string overtones produced by electromagnetically driving piano strings — 44.1 kHz.
+- `crozzoli_bigensemblesmusic_18d.ts` — **Big Ensemble Music (Crozzoli)**. Big-ensemble orchestral music (M. Crozzoli). Broad 18-dim latent for hugely-textured renders. Sample rate not embedded in filename — defaults to 48 kHz.
+
+### Textures / environment
 - `water_pondbrain_b2048_r48000_z16.ts` — **Water (PondBrain)**. Water / aquatic textures. Treats any input as if it were running through liquid — bubbles, ripples, splashes.
 - `magnets_b2048_r48000_z8.ts` — **Magnets**. Ferromagnetic / electromagnetic resonance textures — metallic hums, distant industrial buzz, magnetized-string ringing.
-- `mrp_strengjavera_b2048_r44100_z16.ts` — **Magnetic Resonator Piano (Strengjavera)**. Magnetic Resonator Piano. Sustained metallic-string overtones produced by electromagnetically driving piano strings — 44.1 kHz.
-- `crozzoli_bigensemblesmusic_18d.ts` — **Big Ensemble Music (Crozzoli)**. Big-ensemble orchestral music (M. Crozzoli). Broad 18-dim latent for hugely-textured renders. Sample rate not embedded in filename — defaults to 48000; override via panel if needed.
 
-Each `.ts` checkpoint is accompanied by a `<stem>.json` sidecar with name, license, sample-rate, latent-dim, source URL, and a one-line description.
+---
+
+## Models — ACIDS public catalog (10 models, mirrored 2026-05-18)
+
+Pulled from the canonical anonymous-download endpoint `https://play.forum.ircam.fr/rave-vst-api/get_model/<slug>`.
+Each `.ts` has a matching `<slug>.json` sidecar in the same schema as the IIL set.
+
+| Slug | Display name | Type | Author | Year | Size | Prior |
+|---|---|---|---|---|---|---|
+| `VCTK` | VCTK (English Speech) | RAVE v1 (default) | Jb Dupuy | 2022 | 177 MB | ✓ |
+| `darbouka_onnx` | Darbouka (Percussion) | RAVE v2 (ONNX) | Antoine Caillon | 2022 | 26 MB | – |
+| `nasa` | NASA Apollo 11 | RAVE v1 (default) | Antoine Caillon | 2022 | 159 MB | ✓ |
+| `percussion` | Percussion (Mixed) | RAVE v1 (default) | Antoine Caillon | 2022 | 71 MB | ✓ |
+| `vintage` | Vintage Music | RAVE v1 (large) | Antoine Caillon | 2022 | 482 MB | ✓ |
+| `isis` | ISiS (IRCAM Vocal DB) | RAVE v2 | A. Chemla–Romeu-Santos | 2023 | 149 MB | – |
+| `musicnet` | MusicNet (Classical) | RAVE v2 | A. Chemla–Romeu-Santos | 2023 | 237 MB | ✓ |
+| `sol_ordinario` | Studio OnLine (Ordinario) | RAVE v2 | A. Chemla–Romeu-Santos | 2023 | 149 MB | – |
+| `sol_full` | Studio OnLine (Full) | RAVE v2 | A. Chemla–Romeu-Santos | 2023 | 149 MB | – |
+| `sol_ordinario_fast` | Studio OnLine (Ordinario, fast) | RAVE v2 (small) | A. Chemla–Romeu-Santos | 2023 | 43 MB | – |
+
+**ACIDS set total: ~1.6 GB across 10 models.**
+
+> Note: `VCTK.ts` (ACIDS v1, 48 kHz, original 2022 release) and `voice_vctk_b2048_r44100_z22.ts`
+> (IIL v2 retrain, 44.1 kHz) are *different* models trained on the same source corpus —
+> keep both for comparison.
+
+---
+
+## File format
+
+Each `*.ts` is a [TorchScript](https://pytorch.org/docs/stable/jit.html) export of the RAVE model,
+streaming-mode (causal convolutions, cached state) — ready for realtime or offline inference.
+
+```python
+import torch
+model = torch.jit.load("vintage.ts")
+# Encode (B, 1, T) → latents
+z = model.encode(audio)
+# Decode latents → audio
+y = model.decode(z)
+```
+
+Models with "Prior available" additionally ship a learned prior that can generate latents
+autoregressively (see the [RAVE repo](https://github.com/acids-ircam/RAVE) for usage).
+
+## Where to find more RAVE models
+
+- [Neutone FX models](https://neutone.ai/fx/models) — community + curated `.nm` files (the Neutone wrapper format).
+- [IRCAM Forum projects](https://forum.ircam.fr/) — individual user-submitted models; many require Forum account.
+- [acids-ircam GitHub releases](https://github.com/acids-ircam/RAVE/releases) — reference checkpoints from the maintainers.
+- [IRCAM RAVE Model Challenge 2025](https://forum.ircam.fr/collections/detail/rave-model-challenge-models/) — 11 prize-winner / submission models gated behind a Forum account.
+
+## Citation
+
+```bibtex
+@inproceedings{caillon2021rave,
+  title={RAVE: A variational autoencoder for fast and high-quality neural audio synthesis},
+  author={Caillon, Antoine and Esling, Philippe},
+  booktitle={arXiv preprint arXiv:2111.05011},
+  year={2021}
+}
+```

VCTK.json

@@ -0,0 +1,12 @@
+{
+  "name": "VCTK (English Speech)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 48000,
+  "latent_dim": null,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/VCTK",
+  "description": "Trained on the VCTK speech corpus (CSTR Edinburgh) \u2014 multi-speaker English. Different from voice_vctk_b2048_r44100_z22.ts in this same repo (which is the IIL-curated v2 retrain); this is the original ACIDS v1 release.",
+  "author": "Jb Dupuy",
+  "release_date": "2022-05-11",
+  "model_type": "RAVE v1 (default)",
+  "prior_available": true
+}

VCTK.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d8c8cb4726a2660698de7b6311131975f9f2ddd5915113590b2ae2d9f9c4a38
+size 177167756

darbouka_onnx.json

@@ -0,0 +1,12 @@
+{
+  "name": "Darbouka (Percussion)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 48000,
+  "latent_dim": 4,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/darbouka_onnx",
+  "description": "8 hours of darbouka (goblet drum) recordings \u2014 Middle-Eastern hand drum. ONNX-exported v2. Sharp transients + breathy decays. Good for re-skinning any percussive input into Mediterranean / North-African textures.",
+  "author": "Antoine Caillon",
+  "release_date": "2022-09-21",
+  "model_type": "RAVE v2 (ONNX)",
+  "prior_available": false
+}

darbouka_onnx.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2631209242a90aaeca4c1c2985b1b104595c60713b44fef0a1c1e49b8e995f98
+size 26350036

isis.json

@@ -0,0 +1,12 @@
+{
+  "name": "ISiS (IRCAM Vocal Database)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 44100,
+  "latent_dim": 8,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/isis",
+  "description": "Trained on the IRCAM ISiS vocal analysis-synthesis database. Operatic + extended-technique vocal timbres with rich formant structure. See forum.ircam.fr/projects/detail/isis/ for the source corpus.",
+  "author": "Axel Chemla\u2013Romeu-Santos",
+  "release_date": "2023-12-10",
+  "model_type": "RAVE v2",
+  "prior_available": false
+}

isis.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16c5502112eb27cd9348690a776e1d60ee4f9002e2855d39354d75401b8c408b
+size 148725070

musicnet.json

@@ -0,0 +1,12 @@
+{
+  "name": "MusicNet (Classical Ensemble)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 44100,
+  "latent_dim": 16,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/musicnet",
+  "description": "Trained on the MusicNet database (annotated classical chamber + orchestral recordings). Broad 16-dim latent captures multi-instrument timbral blends; good 'classical-ensemble' transfer target.",
+  "author": "Axel Chemla\u2013Romeu-Santos",
+  "release_date": "2023-12-10",
+  "model_type": "RAVE v2",
+  "prior_available": true
+}

musicnet.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5228bc514c0a38280b774fd5a75ba5883e40013419a5acb0840ebd56d36e855b
+size 236989418

nasa.json

@@ -0,0 +1,12 @@
+{
+  "name": "NASA Apollo 11",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 48000,
+  "latent_dim": 8,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/nasa",
+  "description": "Trained on radio communications from the Apollo 11 mission. Distinctive comms-filtered voice + cosmic interference textures. Use any input \u2192 eerie spacefaring transmissions. Source: youtube.com/watch?v=DejhGSEu8wk",
+  "author": "Antoine Caillon",
+  "release_date": "2022-09-21",
+  "model_type": "RAVE v1 (default)",
+  "prior_available": true
+}

nasa.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60b198ad1ae9450f16c00c6a27ca9332924eb15e57c1a1efb90909038dcde458
+size 159284126

percussion.json

@@ -0,0 +1,12 @@
+{
+  "name": "Percussion (Mixed)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 44100,
+  "latent_dim": null,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/percussion",
+  "description": "8 hours of mixed percussion recordings \u2014 wide range of struck / hit / shaken sources. Good general-purpose 'turn this into rhythm' transfer; complements darbouka with broader timbral coverage.",
+  "author": "Antoine Caillon",
+  "release_date": "2022-09-21",
+  "model_type": "RAVE v1 (default)",
+  "prior_available": true
+}

percussion.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12e044fa4cf0f461fa78dc5e108253d5ffaa395604a66aa43e3a9c3967dee0f9
+size 71097032

sol_full.json

@@ -0,0 +1,12 @@
+{
+  "name": "Studio OnLine (Full)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 44100,
+  "latent_dim": 8,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/sol_full",
+  "description": "Full Studio OnLine database \u2014 ordinario PLUS extended techniques across every catalogued instrument. Bigger sonic palette than sol_ordinario at the cost of less consistency.",
+  "author": "Axel Chemla\u2013Romeu-Santos",
+  "release_date": "2023-12-10",
+  "model_type": "RAVE v2",
+  "prior_available": false
+}

sol_full.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba344fbd634937ed72779264d6d30fb85b9e125e7e1ab687c3c8145902d4a94
+size 148724592

sol_ordinario.json

@@ -0,0 +1,12 @@
+{
+  "name": "Studio OnLine (Ordinario)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 44100,
+  "latent_dim": 4,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/sol_ordinario",
+  "description": "Ordinario (standard playing-technique) recordings from IRCAM's Studio OnLine database \u2014 strings, winds, brass, percussion in conventional articulations. Compact 4-dim latent; very clean re-voicing.",
+  "author": "Axel Chemla\u2013Romeu-Santos",
+  "release_date": "2023-12-10",
+  "model_type": "RAVE v2",
+  "prior_available": false
+}

sol_ordinario.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84a620c61b8d3a28ee35c4abdc796a53bdd17c97adf81e3e4c7243cf8894a3ac
+size 148733766

sol_ordinario_fast.json

@@ -0,0 +1,12 @@
+{
+  "name": "Studio OnLine (Ordinario, fast)",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 44100,
+  "latent_dim": 8,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/sol_ordinario_fast",
+  "description": "Smaller / faster variant of sol_ordinario (43 MB vs 149 MB). Lower-fidelity rendering but much cheaper on CPU. Good default starter pick for live exploration before committing to the larger ordinario model.",
+  "author": "Axel Chemla\u2013Romeu-Santos",
+  "release_date": "2023-10-10",
+  "model_type": "RAVE v2 (small)",
+  "prior_available": false
+}

sol_ordinario_fast.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a8d32002efd6f385b521d64691e0d2d90d691ca4c50e85149ea88c6be399793
+size 43058211

vintage.json

@@ -0,0 +1,12 @@
+{
+  "name": "Vintage Music",
+  "license": "CC-BY-NC-4.0",
+  "sample_rate": 48000,
+  "latent_dim": null,
+  "source_url": "https://play.forum.ircam.fr/rave-vst-api/get_model/vintage",
+  "description": "80 hours of vintage-music recordings \u2014 the canonical 'large' ACIDS demo checkpoint (482 MB). Lush, lo-fi, archival-feeling outputs. Slowest of the set on CPU but the most musically-coherent timbre transfers.",
+  "author": "Antoine Caillon",
+  "release_date": "2022-09-21",
+  "model_type": "RAVE v1 (large)",
+  "prior_available": true
+}

vintage.ts

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba4bb7e281658f2de0e97d47ecbdb7793169804bd1833859619f528b128ebf15
+size 481781124