Spaces:
Sleeping
Sleeping
Hugo Flores Garcia
commited on
Commit
•
bccd2ea
1
Parent(s):
3419098
add stuff
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- .gitignore +190 -0
- LICENSE +21 -0
- README.md +113 -6
- app.py +678 -0
- assets/example.wav +0 -0
- conf/c2f.yml +14 -0
- conf/generated/bbc-humans/c2f.yml +15 -0
- conf/generated/bbc-humans/coarse.yml +8 -0
- conf/generated/bbc-humans/interface.yml +6 -0
- conf/generated/boleros/c2f.yml +15 -0
- conf/generated/boleros/coarse.yml +8 -0
- conf/generated/boleros/interface.yml +6 -0
- conf/generated/bowl/c2f.yml +16 -0
- conf/generated/bowl/coarse.yml +9 -0
- conf/generated/bowl/interface.yml +7 -0
- conf/generated/breaks-steps/interface.yml +8 -0
- conf/generated/choir/interface.yml +9 -0
- conf/generated/earlymachines/c2f.yml +15 -0
- conf/generated/earlymachines/coarse.yml +8 -0
- conf/generated/earlymachines/interface.yml +8 -0
- conf/generated/funk/c2f.yml +15 -0
- conf/generated/funk/coarse.yml +8 -0
- conf/generated/funk/interface.yml +8 -0
- conf/generated/ismir-birds/c2f.yml +15 -0
- conf/generated/ismir-birds/coarse.yml +8 -0
- conf/generated/ismir-birds/interface.yml +8 -0
- conf/generated/ismir-machines/c2f.yml +15 -0
- conf/generated/ismir-machines/coarse.yml +8 -0
- conf/generated/ismir-machines/interface.yml +8 -0
- conf/generated/machines/c2f.yml +15 -0
- conf/generated/machines/coarse.yml +8 -0
- conf/generated/machines/interface.yml +8 -0
- conf/generated/musdb/c2f.yml +40 -0
- conf/generated/musdb/coarse.yml +31 -0
- conf/generated/musdb/interface.yml +8 -0
- conf/generated/n64/c2f.yml +15 -0
- conf/generated/n64/coarse.yml +8 -0
- conf/generated/n64/interface.yml +6 -0
- conf/generated/natural-sounds/c2f.yml +28 -0
- conf/generated/natural-sounds/coarse.yml +21 -0
- conf/generated/natural-sounds/interface.yml +7 -0
- conf/generated/nes/c2f.yml +15 -0
- conf/generated/nes/coarse.yml +8 -0
- conf/generated/nes/interface.yml +6 -0
- conf/generated/nyc-subway/c2f.yml +15 -0
- conf/generated/nyc-subway/coarse.yml +8 -0
- conf/generated/nyc-subway/interface.yml +8 -0
- conf/generated/ocean-waves/c2f.yml +15 -0
- conf/generated/ocean-waves/coarse.yml +8 -0
- conf/generated/ocean-waves/interface.yml +8 -0
.gitignore
ADDED
@@ -0,0 +1,190 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Byte-compiled / optimized / DLL files
|
2 |
+
__pycache__/
|
3 |
+
*.py[cod]
|
4 |
+
*$py.class
|
5 |
+
|
6 |
+
# C extensions
|
7 |
+
*.so
|
8 |
+
|
9 |
+
# Distribution / packaging
|
10 |
+
.Python
|
11 |
+
build/
|
12 |
+
develop-eggs/
|
13 |
+
dist/
|
14 |
+
downloads/
|
15 |
+
eggs/
|
16 |
+
.eggs/
|
17 |
+
lib/
|
18 |
+
lib64/
|
19 |
+
parts/
|
20 |
+
sdist/
|
21 |
+
var/
|
22 |
+
wheels/
|
23 |
+
pip-wheel-metadata/
|
24 |
+
share/python-wheels/
|
25 |
+
*.egg-info/
|
26 |
+
.installed.cfg
|
27 |
+
*.egg
|
28 |
+
MANIFEST
|
29 |
+
|
30 |
+
# PyInstaller
|
31 |
+
# Usually these files are written by a python script from a template
|
32 |
+
# before PyInstaller builds the exe, so as to inject date/other infos into it.
|
33 |
+
*.manifest
|
34 |
+
*.spec
|
35 |
+
|
36 |
+
# Installer logs
|
37 |
+
pip-log.txt
|
38 |
+
pip-delete-this-directory.txt
|
39 |
+
|
40 |
+
# Unit test / coverage reports
|
41 |
+
htmlcov/
|
42 |
+
.tox/
|
43 |
+
.nox/
|
44 |
+
.coverage
|
45 |
+
.coverage.*
|
46 |
+
.cache
|
47 |
+
nosetests.xml
|
48 |
+
coverage.xml
|
49 |
+
*.cover
|
50 |
+
*.py,cover
|
51 |
+
.hypothesis/
|
52 |
+
.pytest_cache/
|
53 |
+
|
54 |
+
# Translations
|
55 |
+
*.mo
|
56 |
+
*.pot
|
57 |
+
|
58 |
+
# Django stuff:
|
59 |
+
*.log
|
60 |
+
local_settings.py
|
61 |
+
db.sqlite3
|
62 |
+
db.sqlite3-journal
|
63 |
+
|
64 |
+
# Flask stuff:
|
65 |
+
instance/
|
66 |
+
.webassets-cache
|
67 |
+
|
68 |
+
# Scrapy stuff:
|
69 |
+
.scrapy
|
70 |
+
|
71 |
+
# Sphinx documentation
|
72 |
+
docs/_build/
|
73 |
+
|
74 |
+
# PyBuilder
|
75 |
+
target/
|
76 |
+
|
77 |
+
# Jupyter Notebook
|
78 |
+
.ipynb_checkpoints
|
79 |
+
|
80 |
+
# IPython
|
81 |
+
profile_default/
|
82 |
+
ipython_config.py
|
83 |
+
|
84 |
+
# pyenv
|
85 |
+
.python-version
|
86 |
+
|
87 |
+
# pipenv
|
88 |
+
# According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
|
89 |
+
# However, in case of collaboration, if having platform-specific dependencies or dependencies
|
90 |
+
# having no cross-platform support, pipenv may install dependencies that don't work, or not
|
91 |
+
# install all needed dependencies.
|
92 |
+
#Pipfile.lock
|
93 |
+
|
94 |
+
# PEP 582; used by e.g. github.com/David-OConnor/pyflow
|
95 |
+
__pypackages__/
|
96 |
+
|
97 |
+
# Celery stuff
|
98 |
+
celerybeat-schedule
|
99 |
+
celerybeat.pid
|
100 |
+
|
101 |
+
# SageMath parsed files
|
102 |
+
*.sage.py
|
103 |
+
|
104 |
+
# Environments
|
105 |
+
.env
|
106 |
+
.venv
|
107 |
+
env/env.sh
|
108 |
+
venv/
|
109 |
+
env.bak/
|
110 |
+
venv.bak/
|
111 |
+
|
112 |
+
# Spyder project settings
|
113 |
+
.spyderproject
|
114 |
+
.spyproject
|
115 |
+
|
116 |
+
# Rope project settings
|
117 |
+
.ropeproject
|
118 |
+
|
119 |
+
# mkdocs documentation
|
120 |
+
/site
|
121 |
+
|
122 |
+
# mypy
|
123 |
+
.mypy_cache/
|
124 |
+
.dmypy.json
|
125 |
+
dmypy.json
|
126 |
+
|
127 |
+
# Pyre type checker
|
128 |
+
.pyre/
|
129 |
+
|
130 |
+
# Files created by experiments
|
131 |
+
output/
|
132 |
+
snapshot/
|
133 |
+
*.m4a
|
134 |
+
notebooks/scratch.ipynb
|
135 |
+
notebooks/inspect.ipynb
|
136 |
+
notebooks/effects.ipynb
|
137 |
+
notebooks/*.ipynb
|
138 |
+
notebooks/*.gif
|
139 |
+
notebooks/*.wav
|
140 |
+
notebooks/*.mp4
|
141 |
+
*runs/
|
142 |
+
boards/
|
143 |
+
samples/
|
144 |
+
*.ipynb
|
145 |
+
|
146 |
+
results.json
|
147 |
+
metrics.csv
|
148 |
+
mprofile_*
|
149 |
+
mem.png
|
150 |
+
|
151 |
+
results/
|
152 |
+
mprofile*
|
153 |
+
*.png
|
154 |
+
# do not ignore the test wav file
|
155 |
+
!tests/audio/short_test_audio.wav
|
156 |
+
!tests/audio/output.wav
|
157 |
+
*/.DS_Store
|
158 |
+
.DS_Store
|
159 |
+
env.sh
|
160 |
+
_codebraid/
|
161 |
+
**/*.html
|
162 |
+
**/*.exec.md
|
163 |
+
flagged/
|
164 |
+
log.txt
|
165 |
+
ckpt/
|
166 |
+
.syncthing*
|
167 |
+
tests/assets/
|
168 |
+
archived/
|
169 |
+
|
170 |
+
scratch/
|
171 |
+
|
172 |
+
runs-archive
|
173 |
+
lyrebird-audiotools
|
174 |
+
lyrebird-audio-codec
|
175 |
+
samples-*/**
|
176 |
+
|
177 |
+
gradio-outputs/
|
178 |
+
samples*/
|
179 |
+
models-all/
|
180 |
+
models.zip
|
181 |
+
.git-old
|
182 |
+
|
183 |
+
|
184 |
+
|
185 |
+
gtzan.zip
|
186 |
+
.gtzan_emb_cache
|
187 |
+
|
188 |
+
|
189 |
+
data/
|
190 |
+
data
|
LICENSE
ADDED
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
MIT License
|
2 |
+
|
3 |
+
Copyright (c) 2023 Hugo Flores García and Prem Seetharaman
|
4 |
+
|
5 |
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
6 |
+
of this software and associated documentation files (the "Software"), to deal
|
7 |
+
in the Software without restriction, including without limitation the rights
|
8 |
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
9 |
+
copies of the Software, and to permit persons to whom the Software is
|
10 |
+
furnished to do so, subject to the following conditions:
|
11 |
+
|
12 |
+
The above copyright notice and this permission notice shall be included in all
|
13 |
+
copies or substantial portions of the Software.
|
14 |
+
|
15 |
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
16 |
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
17 |
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
18 |
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
19 |
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
20 |
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
21 |
+
SOFTWARE.
|
README.md
CHANGED
@@ -1,13 +1,120 @@
|
|
1 |
---
|
2 |
title: Salad Bowl
|
3 |
-
emoji:
|
4 |
-
colorFrom:
|
5 |
-
colorTo:
|
6 |
sdk: gradio
|
7 |
-
sdk_version: 4.
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
-
license: cc-by-nc-
|
11 |
---
|
12 |
|
13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
title: Salad Bowl
|
3 |
+
emoji: 🥗
|
4 |
+
colorFrom: yellow
|
5 |
+
colorTo: green
|
6 |
sdk: gradio
|
7 |
+
sdk_version: 4.11.0
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
+
license: cc-by-nc-4.0
|
11 |
---
|
12 |
|
13 |
+
# VampNet
|
14 |
+
|
15 |
+
This repository contains recipes for training generative music models on top of the Descript Audio Codec.
|
16 |
+
|
17 |
+
## try `unloop`
|
18 |
+
you can try vampnet in a co-creative looper called unloop. see this link: https://github.com/hugofloresgarcia/unloop
|
19 |
+
|
20 |
+
# Setting up
|
21 |
+
|
22 |
+
**Requires Python 3.9**.
|
23 |
+
|
24 |
+
you'll need a Python 3.9 environment to run VampNet. This is due to a [known issue with madmom](https://github.com/hugofloresgarcia/vampnet/issues/15).
|
25 |
+
|
26 |
+
(for example, using conda)
|
27 |
+
```bash
|
28 |
+
conda create -n vampnet python=3.9
|
29 |
+
conda activate vampnet
|
30 |
+
```
|
31 |
+
|
32 |
+
|
33 |
+
install VampNet
|
34 |
+
|
35 |
+
```bash
|
36 |
+
git clone https://github.com/hugofloresgarcia/vampnet.git
|
37 |
+
pip install -e ./vampnet
|
38 |
+
```
|
39 |
+
|
40 |
+
## A note on argbind
|
41 |
+
This repository relies on [argbind](https://github.com/pseeth/argbind) to manage CLIs and config files.
|
42 |
+
Config files are stored in the `conf/` folder.
|
43 |
+
|
44 |
+
## Getting the Pretrained Models
|
45 |
+
|
46 |
+
### Licensing for Pretrained Models:
|
47 |
+
The weights for the models are licensed [`CC BY-NC-SA 4.0`](https://creativecommons.org/licenses/by-nc-sa/4.0/deed.ml). Likewise, any VampNet models fine-tuned on the pretrained models are also licensed [`CC BY-NC-SA 4.0`](https://creativecommons.org/licenses/by-nc-sa/4.0/deed.ml).
|
48 |
+
|
49 |
+
Download the pretrained models from [this link](https://zenodo.org/record/8136629). Then, extract the models to the `models/` folder.
|
50 |
+
|
51 |
+
|
52 |
+
# Usage
|
53 |
+
|
54 |
+
## Launching the Gradio Interface
|
55 |
+
You can launch a gradio UI to play with vampnet.
|
56 |
+
|
57 |
+
```bash
|
58 |
+
python app.py --args.load conf/interface.yml --Interface.device cuda
|
59 |
+
```
|
60 |
+
|
61 |
+
# Training / Fine-tuning
|
62 |
+
|
63 |
+
## Training a model
|
64 |
+
|
65 |
+
To train a model, run the following script:
|
66 |
+
|
67 |
+
```bash
|
68 |
+
python scripts/exp/train.py --args.load conf/vampnet.yml --save_path /path/to/checkpoints
|
69 |
+
```
|
70 |
+
|
71 |
+
for multi-gpu training, use torchrun:
|
72 |
+
|
73 |
+
```bash
|
74 |
+
torchrun --nproc_per_node gpu scripts/exp/train.py --args.load conf/vampnet.yml --save_path path/to/ckpt
|
75 |
+
```
|
76 |
+
|
77 |
+
You can edit `conf/vampnet.yml` to change the dataset paths or any training hyperparameters.
|
78 |
+
|
79 |
+
For coarse2fine models, you can use `conf/c2f.yml` as a starting configuration.
|
80 |
+
|
81 |
+
See `python scripts/exp/train.py -h` for a list of options.
|
82 |
+
|
83 |
+
## Debugging training
|
84 |
+
|
85 |
+
To debug training, it's easier to debug with 1 gpu and 0 workers
|
86 |
+
|
87 |
+
```bash
|
88 |
+
CUDA_VISIBLE_DEVICES=0 python -m pdb scripts/exp/train.py --args.load conf/vampnet.yml --save_path /path/to/checkpoints --num_workers 0
|
89 |
+
```
|
90 |
+
|
91 |
+
## Fine-tuning
|
92 |
+
To fine-tune a model, use the script in `scripts/exp/fine_tune.py` to generate 3 configuration files: `c2f.yml`, `coarse.yml`, and `interface.yml`.
|
93 |
+
The first two are used to fine-tune the coarse and fine models, respectively. The last one is used to launch the gradio interface.
|
94 |
+
|
95 |
+
```bash
|
96 |
+
python scripts/exp/fine_tune.py "/path/to/audio1.mp3 /path/to/audio2/ /path/to/audio3.wav" <fine_tune_name>
|
97 |
+
```
|
98 |
+
|
99 |
+
This will create a folder under `conf/<fine_tune_name>/` with the 3 configuration files.
|
100 |
+
|
101 |
+
The save_paths will be set to `runs/<fine_tune_name>/coarse` and `runs/<fine_tune_name>/c2f`.
|
102 |
+
|
103 |
+
launch the coarse job:
|
104 |
+
```bash
|
105 |
+
python scripts/exp/train.py --args.load conf/generated/<fine_tune_name>/coarse.yml
|
106 |
+
```
|
107 |
+
|
108 |
+
this will save the coarse model to `runs/<fine_tune_name>/coarse/ckpt/best/`.
|
109 |
+
|
110 |
+
launch the c2f job:
|
111 |
+
```bash
|
112 |
+
python scripts/exp/train.py --args.load conf/generated/<fine_tune_name>/c2f.yml
|
113 |
+
```
|
114 |
+
|
115 |
+
launch the interface:
|
116 |
+
```bash
|
117 |
+
python app.py --args.load conf/generated/<fine_tune_name>/interface.yml
|
118 |
+
```
|
119 |
+
|
120 |
+
|
app.py
ADDED
@@ -0,0 +1,678 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from pathlib import Path
|
2 |
+
import yaml
|
3 |
+
import uuid
|
4 |
+
|
5 |
+
import numpy as np
|
6 |
+
import audiotools as at
|
7 |
+
import argbind
|
8 |
+
import shutil
|
9 |
+
import torch
|
10 |
+
|
11 |
+
import gradio as gr
|
12 |
+
from vampnet.interface import Interface
|
13 |
+
from vampnet import mask as pmask
|
14 |
+
|
15 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
16 |
+
|
17 |
+
interface = Interface(
|
18 |
+
device=device,
|
19 |
+
coarse_ckpt="models/vampnet/coarse.pth",
|
20 |
+
coarse2fine_ckpt="models/vampnet/c2f.pth",
|
21 |
+
codec_ckpt="models/vampnet/codec.pth",
|
22 |
+
)
|
23 |
+
|
24 |
+
# populate the model choices with any interface.yml files in the generated confs
|
25 |
+
MODEL_CHOICES = {
|
26 |
+
"default": {
|
27 |
+
"Interface.coarse_ckpt": str(interface.coarse_path),
|
28 |
+
"Interface.coarse2fine_ckpt": str(interface.c2f_path),
|
29 |
+
"Interface.codec_ckpt": str(interface.codec_path),
|
30 |
+
}
|
31 |
+
}
|
32 |
+
generated_confs = Path("conf/generated")
|
33 |
+
for conf_file in generated_confs.glob("*/interface.yml"):
|
34 |
+
with open(conf_file) as f:
|
35 |
+
_conf = yaml.safe_load(f)
|
36 |
+
MODEL_CHOICES[conf_file.parent.name] = _conf
|
37 |
+
|
38 |
+
|
39 |
+
|
40 |
+
OUT_DIR = Path("gradio-outputs")
|
41 |
+
OUT_DIR.mkdir(exist_ok=True, parents=True)
|
42 |
+
|
43 |
+
def load_audio(file):
|
44 |
+
print(file)
|
45 |
+
filepath = file.name
|
46 |
+
sig = at.AudioSignal.salient_excerpt(
|
47 |
+
filepath,
|
48 |
+
duration=interface.coarse.chunk_size_s
|
49 |
+
)
|
50 |
+
sig = interface.preprocess(sig)
|
51 |
+
|
52 |
+
out_dir = OUT_DIR / "tmp" / str(uuid.uuid4())
|
53 |
+
out_dir.mkdir(parents=True, exist_ok=True)
|
54 |
+
sig.write(out_dir / "input.wav")
|
55 |
+
return sig.path_to_file
|
56 |
+
|
57 |
+
|
58 |
+
def load_example_audio():
|
59 |
+
return "./assets/example.wav"
|
60 |
+
|
61 |
+
from torch_pitch_shift import pitch_shift, get_fast_shifts
|
62 |
+
def shift_pitch(signal, interval: int):
|
63 |
+
signal.samples = pitch_shift(
|
64 |
+
signal.samples,
|
65 |
+
shift=interval,
|
66 |
+
sample_rate=signal.sample_rate
|
67 |
+
)
|
68 |
+
return signal
|
69 |
+
|
70 |
+
def _vamp(data, return_mask=False):
|
71 |
+
|
72 |
+
# remove any old files in the output directory (from previous runs)
|
73 |
+
shutil.rmtree(OUT_DIR)
|
74 |
+
OUT_DIR.mkdir()
|
75 |
+
|
76 |
+
out_dir = OUT_DIR / str(uuid.uuid4())
|
77 |
+
out_dir.mkdir()
|
78 |
+
|
79 |
+
sig = at.AudioSignal(data[input_audio])
|
80 |
+
sig = interface.preprocess(sig)
|
81 |
+
|
82 |
+
|
83 |
+
# reload the model if necessary
|
84 |
+
interface.reload(
|
85 |
+
coarse_ckpt=MODEL_CHOICES[data[model_choice]]["Interface.coarse_ckpt"],
|
86 |
+
c2f_ckpt=MODEL_CHOICES[data[model_choice]]["Interface.coarse2fine_ckpt"],
|
87 |
+
)
|
88 |
+
|
89 |
+
loudness = sig.loudness()
|
90 |
+
print(f"input loudness is {loudness}")
|
91 |
+
|
92 |
+
if data[pitch_shift_amt] != 0:
|
93 |
+
sig = shift_pitch(sig, data[pitch_shift_amt])
|
94 |
+
|
95 |
+
z = interface.encode(sig)
|
96 |
+
|
97 |
+
ncc = data[n_conditioning_codebooks]
|
98 |
+
|
99 |
+
# build the mask
|
100 |
+
mask = pmask.linear_random(z, data[rand_mask_intensity])
|
101 |
+
mask = pmask.mask_and(
|
102 |
+
mask, pmask.inpaint(
|
103 |
+
z,
|
104 |
+
interface.s2t(data[prefix_s]),
|
105 |
+
interface.s2t(data[suffix_s])
|
106 |
+
)
|
107 |
+
)
|
108 |
+
mask = pmask.mask_and(
|
109 |
+
mask, pmask.periodic_mask(
|
110 |
+
z,
|
111 |
+
data[periodic_p],
|
112 |
+
data[periodic_w],
|
113 |
+
random_roll=True
|
114 |
+
)
|
115 |
+
)
|
116 |
+
if data[onset_mask_width] > 0:
|
117 |
+
mask = pmask.mask_or(
|
118 |
+
mask, pmask.onset_mask(sig, z, interface, width=data[onset_mask_width])
|
119 |
+
)
|
120 |
+
if data[beat_mask_width] > 0:
|
121 |
+
beat_mask = interface.make_beat_mask(
|
122 |
+
sig,
|
123 |
+
after_beat_s=(data[beat_mask_width]/1000),
|
124 |
+
mask_upbeats=not data[beat_mask_downbeats],
|
125 |
+
)
|
126 |
+
mask = pmask.mask_and(mask, beat_mask)
|
127 |
+
|
128 |
+
# these should be the last two mask ops
|
129 |
+
mask = pmask.dropout(mask, data[dropout])
|
130 |
+
mask = pmask.codebook_unmask(mask, ncc)
|
131 |
+
mask = pmask.codebook_mask(mask, int(data[n_mask_codebooks]))
|
132 |
+
|
133 |
+
print(f"dropout {data[dropout]}")
|
134 |
+
print(f"masktemp {data[masktemp]}")
|
135 |
+
print(f"sampletemp {data[sampletemp]}")
|
136 |
+
print(f"top_p {data[top_p]}")
|
137 |
+
print(f"prefix_s {data[prefix_s]}")
|
138 |
+
print(f"suffix_s {data[suffix_s]}")
|
139 |
+
print(f"rand_mask_intensity {data[rand_mask_intensity]}")
|
140 |
+
print(f"num_steps {data[num_steps]}")
|
141 |
+
print(f"periodic_p {data[periodic_p]}")
|
142 |
+
print(f"periodic_w {data[periodic_w]}")
|
143 |
+
print(f"n_conditioning_codebooks {data[n_conditioning_codebooks]}")
|
144 |
+
print(f"use_coarse2fine {data[use_coarse2fine]}")
|
145 |
+
print(f"onset_mask_width {data[onset_mask_width]}")
|
146 |
+
print(f"beat_mask_width {data[beat_mask_width]}")
|
147 |
+
print(f"beat_mask_downbeats {data[beat_mask_downbeats]}")
|
148 |
+
print(f"stretch_factor {data[stretch_factor]}")
|
149 |
+
print(f"seed {data[seed]}")
|
150 |
+
print(f"pitch_shift_amt {data[pitch_shift_amt]}")
|
151 |
+
print(f"sample_cutoff {data[sample_cutoff]}")
|
152 |
+
|
153 |
+
|
154 |
+
_top_p = data[top_p] if data[top_p] > 0 else None
|
155 |
+
# save the mask as a txt file
|
156 |
+
np.savetxt(out_dir / "mask.txt", mask[:,0,:].long().cpu().numpy())
|
157 |
+
|
158 |
+
_seed = data[seed] if data[seed] > 0 else None
|
159 |
+
print(f"processing coarse...")
|
160 |
+
zv, mask_z = interface.coarse_vamp(
|
161 |
+
z,
|
162 |
+
mask=mask,
|
163 |
+
sampling_steps=data[num_steps],
|
164 |
+
mask_temperature=data[masktemp]*10,
|
165 |
+
sampling_temperature=data[sampletemp],
|
166 |
+
return_mask=True,
|
167 |
+
typical_filtering=data[typical_filtering],
|
168 |
+
typical_mass=data[typical_mass],
|
169 |
+
typical_min_tokens=data[typical_min_tokens],
|
170 |
+
top_p=_top_p,
|
171 |
+
gen_fn=interface.coarse.generate,
|
172 |
+
seed=_seed,
|
173 |
+
sample_cutoff=data[sample_cutoff],
|
174 |
+
)
|
175 |
+
|
176 |
+
if use_coarse2fine:
|
177 |
+
print(f"processing coarse to fine...")
|
178 |
+
zv = interface.coarse_to_fine(
|
179 |
+
zv,
|
180 |
+
mask_temperature=data[masktemp]*10,
|
181 |
+
sampling_temperature=data[sampletemp],
|
182 |
+
mask=mask,
|
183 |
+
sampling_steps=data[num_steps] // 2,
|
184 |
+
sample_cutoff=data[sample_cutoff],
|
185 |
+
seed=_seed,
|
186 |
+
)
|
187 |
+
|
188 |
+
sig = interface.to_signal(zv).cpu()
|
189 |
+
print("done")
|
190 |
+
|
191 |
+
print(f"output loudness is {sig.loudness()}")
|
192 |
+
sig = sig.normalize(loudness)
|
193 |
+
print(f"normalized loudness is {sig.loudness()}")
|
194 |
+
print("\n")
|
195 |
+
|
196 |
+
sig.write(out_dir / "output.wav")
|
197 |
+
|
198 |
+
if return_mask:
|
199 |
+
mask = interface.to_signal(mask_z).cpu()
|
200 |
+
mask.write(out_dir / "mask.wav")
|
201 |
+
return sig.path_to_file, mask.path_to_file
|
202 |
+
else:
|
203 |
+
return sig.path_to_file
|
204 |
+
|
205 |
+
def vamp(data):
|
206 |
+
return _vamp(data, return_mask=True)
|
207 |
+
|
208 |
+
def api_vamp(data):
|
209 |
+
return _vamp(data, return_mask=False)
|
210 |
+
|
211 |
+
def save_vamp(data):
|
212 |
+
out_dir = OUT_DIR / "saved" / str(uuid.uuid4())
|
213 |
+
out_dir.mkdir(parents=True, exist_ok=True)
|
214 |
+
|
215 |
+
sig_in = at.AudioSignal(data[input_audio])
|
216 |
+
sig_out = at.AudioSignal(data[output_audio])
|
217 |
+
|
218 |
+
sig_in.write(out_dir / "input.wav")
|
219 |
+
sig_out.write(out_dir / "output.wav")
|
220 |
+
|
221 |
+
_data = {
|
222 |
+
"masktemp": data[masktemp],
|
223 |
+
"sampletemp": data[sampletemp],
|
224 |
+
"top_p": data[top_p],
|
225 |
+
"prefix_s": data[prefix_s],
|
226 |
+
"suffix_s": data[suffix_s],
|
227 |
+
"rand_mask_intensity": data[rand_mask_intensity],
|
228 |
+
"num_steps": data[num_steps],
|
229 |
+
"notes": data[notes_text],
|
230 |
+
"periodic_period": data[periodic_p],
|
231 |
+
"periodic_width": data[periodic_w],
|
232 |
+
"n_conditioning_codebooks": data[n_conditioning_codebooks],
|
233 |
+
"use_coarse2fine": data[use_coarse2fine],
|
234 |
+
"stretch_factor": data[stretch_factor],
|
235 |
+
"seed": data[seed],
|
236 |
+
"samplecutoff": data[sample_cutoff],
|
237 |
+
}
|
238 |
+
|
239 |
+
# save with yaml
|
240 |
+
with open(out_dir / "data.yaml", "w") as f:
|
241 |
+
yaml.dump(_data, f)
|
242 |
+
|
243 |
+
import zipfile
|
244 |
+
zip_path = str(out_dir.with_suffix(".zip"))
|
245 |
+
with zipfile.ZipFile(zip_path, "w") as zf:
|
246 |
+
for file in out_dir.iterdir():
|
247 |
+
zf.write(file, file.name)
|
248 |
+
|
249 |
+
return f"saved! your save code is {out_dir.stem}", zip_path
|
250 |
+
|
251 |
+
|
252 |
+
def harp_vamp(_input_audio, _beat_mask_width, _sampletemp):
|
253 |
+
|
254 |
+
out_dir = OUT_DIR / str(uuid.uuid4())
|
255 |
+
out_dir.mkdir()
|
256 |
+
sig = at.AudioSignal(_input_audio)
|
257 |
+
sig = interface.preprocess(sig)
|
258 |
+
|
259 |
+
z = interface.encode(sig)
|
260 |
+
|
261 |
+
# build the mask
|
262 |
+
mask = pmask.linear_random(z, 1.0)
|
263 |
+
if _beat_mask_width > 0:
|
264 |
+
beat_mask = interface.make_beat_mask(
|
265 |
+
sig,
|
266 |
+
after_beat_s=(_beat_mask_width/1000),
|
267 |
+
)
|
268 |
+
mask = pmask.mask_and(mask, beat_mask)
|
269 |
+
|
270 |
+
# save the mask as a txt file
|
271 |
+
zv, mask_z = interface.coarse_vamp(
|
272 |
+
z,
|
273 |
+
mask=mask,
|
274 |
+
sampling_temperature=_sampletemp,
|
275 |
+
return_mask=True,
|
276 |
+
gen_fn=interface.coarse.generate,
|
277 |
+
)
|
278 |
+
|
279 |
+
|
280 |
+
zv = interface.coarse_to_fine(
|
281 |
+
zv,
|
282 |
+
sampling_temperature=_sampletemp,
|
283 |
+
mask=mask,
|
284 |
+
)
|
285 |
+
|
286 |
+
sig = interface.to_signal(zv).cpu()
|
287 |
+
print("done")
|
288 |
+
|
289 |
+
sig.write(out_dir / "output.wav")
|
290 |
+
|
291 |
+
return sig.path_to_file
|
292 |
+
|
293 |
+
with gr.Blocks() as demo:
|
294 |
+
|
295 |
+
with gr.Row():
|
296 |
+
with gr.Column():
|
297 |
+
gr.Markdown("# VampNet Audio Vamping")
|
298 |
+
gr.Markdown("""## Description:
|
299 |
+
This is a demo of the VampNet, a generative audio model that transforms the input audio based on the chosen settings.
|
300 |
+
You can control the extent and nature of variation with a set of manual controls and presets.
|
301 |
+
Use this interface to experiment with different mask settings and explore the audio outputs.
|
302 |
+
""")
|
303 |
+
|
304 |
+
gr.Markdown("""
|
305 |
+
## Instructions:
|
306 |
+
1. You can start by uploading some audio, or by loading the example audio.
|
307 |
+
2. Choose a preset for the vamp operation, or manually adjust the controls to customize the mask settings.
|
308 |
+
3. Click the "generate (vamp)!!!" button to apply the vamp operation. Listen to the output audio.
|
309 |
+
4. Optionally, you can add some notes and save the result.
|
310 |
+
5. You can also use the output as the new input and continue experimenting!
|
311 |
+
""")
|
312 |
+
with gr.Row():
|
313 |
+
with gr.Column():
|
314 |
+
|
315 |
+
|
316 |
+
manual_audio_upload = gr.File(
|
317 |
+
label=f"upload some audio (will be randomly trimmed to max of {interface.coarse.chunk_size_s:.2f}s)",
|
318 |
+
file_types=["audio"]
|
319 |
+
)
|
320 |
+
load_example_audio_button = gr.Button("or load example audio")
|
321 |
+
|
322 |
+
input_audio = gr.Audio(
|
323 |
+
label="input audio",
|
324 |
+
interactive=False,
|
325 |
+
type="filepath",
|
326 |
+
)
|
327 |
+
|
328 |
+
audio_mask = gr.Audio(
|
329 |
+
label="audio mask (listen to this to hear the mask hints)",
|
330 |
+
interactive=False,
|
331 |
+
type="filepath",
|
332 |
+
)
|
333 |
+
|
334 |
+
# connect widgets
|
335 |
+
load_example_audio_button.click(
|
336 |
+
fn=load_example_audio,
|
337 |
+
inputs=[],
|
338 |
+
outputs=[ input_audio]
|
339 |
+
)
|
340 |
+
|
341 |
+
manual_audio_upload.change(
|
342 |
+
fn=load_audio,
|
343 |
+
inputs=[manual_audio_upload],
|
344 |
+
outputs=[ input_audio]
|
345 |
+
)
|
346 |
+
|
347 |
+
# mask settings
|
348 |
+
with gr.Column():
|
349 |
+
|
350 |
+
|
351 |
+
presets = {
|
352 |
+
"unconditional": {
|
353 |
+
"periodic_p": 0,
|
354 |
+
"onset_mask_width": 0,
|
355 |
+
"beat_mask_width": 0,
|
356 |
+
"beat_mask_downbeats": False,
|
357 |
+
},
|
358 |
+
"slight periodic variation": {
|
359 |
+
"periodic_p": 5,
|
360 |
+
"onset_mask_width": 5,
|
361 |
+
"beat_mask_width": 0,
|
362 |
+
"beat_mask_downbeats": False,
|
363 |
+
},
|
364 |
+
"moderate periodic variation": {
|
365 |
+
"periodic_p": 13,
|
366 |
+
"onset_mask_width": 5,
|
367 |
+
"beat_mask_width": 0,
|
368 |
+
"beat_mask_downbeats": False,
|
369 |
+
},
|
370 |
+
"strong periodic variation": {
|
371 |
+
"periodic_p": 17,
|
372 |
+
"onset_mask_width": 5,
|
373 |
+
"beat_mask_width": 0,
|
374 |
+
"beat_mask_downbeats": False,
|
375 |
+
},
|
376 |
+
"very strong periodic variation": {
|
377 |
+
"periodic_p": 21,
|
378 |
+
"onset_mask_width": 5,
|
379 |
+
"beat_mask_width": 0,
|
380 |
+
"beat_mask_downbeats": False,
|
381 |
+
},
|
382 |
+
"beat-driven variation": {
|
383 |
+
"periodic_p": 0,
|
384 |
+
"onset_mask_width": 0,
|
385 |
+
"beat_mask_width": 50,
|
386 |
+
"beat_mask_downbeats": False,
|
387 |
+
},
|
388 |
+
"beat-driven variation (downbeats only)": {
|
389 |
+
"periodic_p": 0,
|
390 |
+
"onset_mask_width": 0,
|
391 |
+
"beat_mask_width": 50,
|
392 |
+
"beat_mask_downbeats": True,
|
393 |
+
},
|
394 |
+
"beat-driven variation (downbeats only, strong)": {
|
395 |
+
"periodic_p": 0,
|
396 |
+
"onset_mask_width": 0,
|
397 |
+
"beat_mask_width": 20,
|
398 |
+
"beat_mask_downbeats": True,
|
399 |
+
},
|
400 |
+
}
|
401 |
+
|
402 |
+
preset = gr.Dropdown(
|
403 |
+
label="preset",
|
404 |
+
choices=list(presets.keys()),
|
405 |
+
value="strong periodic variation",
|
406 |
+
)
|
407 |
+
load_preset_button = gr.Button("load_preset")
|
408 |
+
|
409 |
+
with gr.Accordion("manual controls", open=True):
|
410 |
+
periodic_p = gr.Slider(
|
411 |
+
label="periodic prompt (0 - unconditional, 2 - lots of hints, 8 - a couple of hints, 16 - occasional hint, 32 - very occasional hint, etc)",
|
412 |
+
minimum=0,
|
413 |
+
maximum=128,
|
414 |
+
step=1,
|
415 |
+
value=3,
|
416 |
+
)
|
417 |
+
|
418 |
+
|
419 |
+
onset_mask_width = gr.Slider(
|
420 |
+
label="onset mask width (multiplies with the periodic mask, 1 step ~= 10milliseconds) ",
|
421 |
+
minimum=0,
|
422 |
+
maximum=100,
|
423 |
+
step=1,
|
424 |
+
value=5,
|
425 |
+
)
|
426 |
+
|
427 |
+
beat_mask_width = gr.Slider(
|
428 |
+
label="beat prompt (ms)",
|
429 |
+
minimum=0,
|
430 |
+
maximum=200,
|
431 |
+
value=0,
|
432 |
+
)
|
433 |
+
beat_mask_downbeats = gr.Checkbox(
|
434 |
+
label="beat mask downbeats only?",
|
435 |
+
value=False
|
436 |
+
)
|
437 |
+
|
438 |
+
n_mask_codebooks = gr.Number(
|
439 |
+
label="first upper codebook level to mask",
|
440 |
+
value=9,
|
441 |
+
)
|
442 |
+
|
443 |
+
|
444 |
+
with gr.Accordion("extras ", open=False):
|
445 |
+
pitch_shift_amt = gr.Slider(
|
446 |
+
label="pitch shift amount (semitones)",
|
447 |
+
minimum=-12,
|
448 |
+
maximum=12,
|
449 |
+
step=1,
|
450 |
+
value=0,
|
451 |
+
)
|
452 |
+
|
453 |
+
rand_mask_intensity = gr.Slider(
|
454 |
+
label="random mask intensity. (If this is less than 1, scatters prompts throughout the audio, should be between 0.9 and 1.0)",
|
455 |
+
minimum=0.0,
|
456 |
+
maximum=1.0,
|
457 |
+
value=1.0
|
458 |
+
)
|
459 |
+
|
460 |
+
periodic_w = gr.Slider(
|
461 |
+
label="periodic prompt width (steps, 1 step ~= 10milliseconds)",
|
462 |
+
minimum=1,
|
463 |
+
maximum=20,
|
464 |
+
step=1,
|
465 |
+
value=1,
|
466 |
+
)
|
467 |
+
n_conditioning_codebooks = gr.Number(
|
468 |
+
label="number of conditioning codebooks. probably 0",
|
469 |
+
value=0,
|
470 |
+
precision=0,
|
471 |
+
)
|
472 |
+
|
473 |
+
stretch_factor = gr.Slider(
|
474 |
+
label="time stretch factor",
|
475 |
+
minimum=0,
|
476 |
+
maximum=64,
|
477 |
+
step=1,
|
478 |
+
value=1,
|
479 |
+
)
|
480 |
+
|
481 |
+
preset_outputs = {
|
482 |
+
periodic_p,
|
483 |
+
onset_mask_width,
|
484 |
+
beat_mask_width,
|
485 |
+
beat_mask_downbeats,
|
486 |
+
}
|
487 |
+
|
488 |
+
def load_preset(_preset):
|
489 |
+
return tuple(presets[_preset].values())
|
490 |
+
|
491 |
+
load_preset_button.click(
|
492 |
+
fn=load_preset,
|
493 |
+
inputs=[preset],
|
494 |
+
outputs=preset_outputs
|
495 |
+
)
|
496 |
+
|
497 |
+
|
498 |
+
with gr.Accordion("prefix/suffix prompts", open=False):
|
499 |
+
prefix_s = gr.Slider(
|
500 |
+
label="prefix hint length (seconds)",
|
501 |
+
minimum=0.0,
|
502 |
+
maximum=10.0,
|
503 |
+
value=0.0
|
504 |
+
)
|
505 |
+
suffix_s = gr.Slider(
|
506 |
+
label="suffix hint length (seconds)",
|
507 |
+
minimum=0.0,
|
508 |
+
maximum=10.0,
|
509 |
+
value=0.0
|
510 |
+
)
|
511 |
+
|
512 |
+
masktemp = gr.Slider(
|
513 |
+
label="mask temperature",
|
514 |
+
minimum=0.0,
|
515 |
+
maximum=100.0,
|
516 |
+
value=1.5
|
517 |
+
)
|
518 |
+
sampletemp = gr.Slider(
|
519 |
+
label="sample temperature",
|
520 |
+
minimum=0.1,
|
521 |
+
maximum=10.0,
|
522 |
+
value=1.0,
|
523 |
+
step=0.001
|
524 |
+
)
|
525 |
+
|
526 |
+
|
527 |
+
|
528 |
+
with gr.Accordion("sampling settings", open=False):
|
529 |
+
top_p = gr.Slider(
|
530 |
+
label="top p (0.0 = off)",
|
531 |
+
minimum=0.0,
|
532 |
+
maximum=1.0,
|
533 |
+
value=0.0
|
534 |
+
)
|
535 |
+
typical_filtering = gr.Checkbox(
|
536 |
+
label="typical filtering ",
|
537 |
+
value=False
|
538 |
+
)
|
539 |
+
typical_mass = gr.Slider(
|
540 |
+
label="typical mass (should probably stay between 0.1 and 0.5)",
|
541 |
+
minimum=0.01,
|
542 |
+
maximum=0.99,
|
543 |
+
value=0.15
|
544 |
+
)
|
545 |
+
typical_min_tokens = gr.Slider(
|
546 |
+
label="typical min tokens (should probably stay between 1 and 256)",
|
547 |
+
minimum=1,
|
548 |
+
maximum=256,
|
549 |
+
step=1,
|
550 |
+
value=64
|
551 |
+
)
|
552 |
+
sample_cutoff = gr.Slider(
|
553 |
+
label="sample cutoff",
|
554 |
+
minimum=0.0,
|
555 |
+
maximum=1.0,
|
556 |
+
value=0.5,
|
557 |
+
step=0.01
|
558 |
+
)
|
559 |
+
|
560 |
+
use_coarse2fine = gr.Checkbox(
|
561 |
+
label="use coarse2fine",
|
562 |
+
value=True,
|
563 |
+
visible=False
|
564 |
+
)
|
565 |
+
|
566 |
+
num_steps = gr.Slider(
|
567 |
+
label="number of steps (should normally be between 12 and 36)",
|
568 |
+
minimum=1,
|
569 |
+
maximum=128,
|
570 |
+
step=1,
|
571 |
+
value=36
|
572 |
+
)
|
573 |
+
|
574 |
+
dropout = gr.Slider(
|
575 |
+
label="mask dropout",
|
576 |
+
minimum=0.0,
|
577 |
+
maximum=1.0,
|
578 |
+
step=0.01,
|
579 |
+
value=0.0
|
580 |
+
)
|
581 |
+
|
582 |
+
|
583 |
+
seed = gr.Number(
|
584 |
+
label="seed (0 for random)",
|
585 |
+
value=0,
|
586 |
+
precision=0,
|
587 |
+
)
|
588 |
+
|
589 |
+
|
590 |
+
|
591 |
+
# mask settings
|
592 |
+
with gr.Column():
|
593 |
+
|
594 |
+
model_choice = gr.Dropdown(
|
595 |
+
label="model choice",
|
596 |
+
choices=list(MODEL_CHOICES.keys()),
|
597 |
+
value="default",
|
598 |
+
visible=True
|
599 |
+
)
|
600 |
+
|
601 |
+
vamp_button = gr.Button("generate (vamp)!!!")
|
602 |
+
output_audio = gr.Audio(
|
603 |
+
label="output audio",
|
604 |
+
interactive=False,
|
605 |
+
type="filepath"
|
606 |
+
)
|
607 |
+
|
608 |
+
notes_text = gr.Textbox(
|
609 |
+
label="type any notes about the generated audio here",
|
610 |
+
value="",
|
611 |
+
interactive=True
|
612 |
+
)
|
613 |
+
save_button = gr.Button("save vamp")
|
614 |
+
download_file = gr.File(
|
615 |
+
label="vamp to download will appear here",
|
616 |
+
interactive=False
|
617 |
+
)
|
618 |
+
use_as_input_button = gr.Button("use output as input")
|
619 |
+
|
620 |
+
thank_you = gr.Markdown("")
|
621 |
+
|
622 |
+
|
623 |
+
_inputs = {
|
624 |
+
input_audio,
|
625 |
+
num_steps,
|
626 |
+
masktemp,
|
627 |
+
sampletemp,
|
628 |
+
top_p,
|
629 |
+
prefix_s, suffix_s,
|
630 |
+
rand_mask_intensity,
|
631 |
+
periodic_p, periodic_w,
|
632 |
+
n_conditioning_codebooks,
|
633 |
+
dropout,
|
634 |
+
use_coarse2fine,
|
635 |
+
stretch_factor,
|
636 |
+
onset_mask_width,
|
637 |
+
typical_filtering,
|
638 |
+
typical_mass,
|
639 |
+
typical_min_tokens,
|
640 |
+
beat_mask_width,
|
641 |
+
beat_mask_downbeats,
|
642 |
+
seed,
|
643 |
+
model_choice,
|
644 |
+
n_mask_codebooks,
|
645 |
+
pitch_shift_amt,
|
646 |
+
sample_cutoff
|
647 |
+
}
|
648 |
+
|
649 |
+
# connect widgets
|
650 |
+
vamp_button.click(
|
651 |
+
fn=vamp,
|
652 |
+
inputs=_inputs,
|
653 |
+
outputs=[output_audio, audio_mask],
|
654 |
+
)
|
655 |
+
|
656 |
+
api_vamp_button = gr.Button("api vamp", visible=False)
|
657 |
+
api_vamp_button.click(
|
658 |
+
fn=api_vamp,
|
659 |
+
inputs=_inputs,
|
660 |
+
outputs=[output_audio],
|
661 |
+
api_name="vamp"
|
662 |
+
)
|
663 |
+
|
664 |
+
use_as_input_button.click(
|
665 |
+
fn=lambda x: x,
|
666 |
+
inputs=[output_audio],
|
667 |
+
outputs=[input_audio]
|
668 |
+
)
|
669 |
+
|
670 |
+
save_button.click(
|
671 |
+
fn=save_vamp,
|
672 |
+
inputs=_inputs | {notes_text, output_audio},
|
673 |
+
outputs=[thank_you, download_file]
|
674 |
+
)
|
675 |
+
|
676 |
+
|
677 |
+
demo.launch(share=True, debug=True)
|
678 |
+
demo.queue()
|
assets/example.wav
ADDED
Binary file (883 kB). View file
|
|
conf/c2f.yml
ADDED
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/vampnet.yml
|
3 |
+
|
4 |
+
VampNet.n_codebooks: 14
|
5 |
+
VampNet.n_conditioning_codebooks: 4
|
6 |
+
|
7 |
+
VampNet.embedding_dim: 1280
|
8 |
+
VampNet.n_layers: 16
|
9 |
+
VampNet.n_heads: 20
|
10 |
+
|
11 |
+
AudioDataset.duration: 3.0
|
12 |
+
|
13 |
+
|
14 |
+
AudioDataset.loudness_cutoff: -40.0
|
conf/generated/bbc-humans/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/bbc-humans/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /home/hugo/Humans/
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/bbc-humans/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/bbc-humans/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /home/hugo/Humans/
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/bbc-humans/interface.yml
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /home/hugo/Humans/
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/bbc-humans/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse_ckpt: ./runs/bbc-humans/coarse/latest/vampnet/weights.pth
|
5 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
6 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/boleros/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/boleros/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/loras/boleros
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/boleros/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/boleros/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/loras/boleros
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/boleros/interface.yml
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/loras/boleros
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/boleros/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse_ckpt: ./runs/boleros/coarse/latest/vampnet/weights.pth
|
5 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
6 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/bowl/c2f.yml
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/bowl/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/seagate_prosound/prosound_core_complete/Anns
|
15 |
+
- Animals
|
16 |
+
val/AudioLoader.sources: *id001
|
conf/generated/bowl/coarse.yml
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/bowl/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/seagate_prosound/prosound_core_complete/Anns
|
8 |
+
- Animals
|
9 |
+
val/AudioLoader.sources: *id001
|
conf/generated/bowl/interface.yml
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/seagate_prosound/prosound_core_complete/Anns
|
3 |
+
- Animals
|
4 |
+
Interface.coarse2fine_ckpt: ./runs/bowl/c2f/latest/vampnet/weights.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/bowl/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
7 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/breaks-steps/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/machines
|
3 |
+
Interface.coarse2fine_ckpt: ./runs-june-23/breaks-steps/c2f/best/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs-june-23/breaks-steps/c2f/best/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs-june-23/breaks-steps/coarse/best/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs-june-23/breaks-steps/coarse/best/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/choir/interface.yml
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/knower
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/choir/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/choir/c2f/latest/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/choir/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/choir/coarse/latest/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
9 |
+
Interface.coarse_chunk_size_s: 15
|
conf/generated/earlymachines/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/machines/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/machines
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/earlymachines/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/machines/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/machines
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/earlymachines/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/machines
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/machines/c2f/best/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/machines/c2f/best/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/machines/coarse/best/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/machines/coarse/best/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/funk/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/knower/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/knower
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/funk/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/knower/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/knower
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/funk/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/knower
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/knower/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/knower/c2f/latest/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/knower/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/knower/coarse/latest/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/ismir-birds/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/ismir-birds/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/ismir-23-cfm/xeno-canto-augmented-filtered
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/ismir-birds/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/ismir-birds/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/ismir-23-cfm/xeno-canto-augmented-filtered
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/ismir-birds/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/ismir-23-cfm/xeno-canto-augmented-filtered
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/ismir-birds/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/ismir-birds/c2f/latest/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/ismir-birds/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/ismir-birds/coarse/latest/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/ismir-machines/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/ismir-machines/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/ismir-23-cfm/machines-augmented-filtered
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/ismir-machines/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/ismir-machines/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/ismir-23-cfm/machines-augmented-filtered
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/ismir-machines/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/ismir-23-cfm/machines-augmented-filtered
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/ismir-machines/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/ismir-machines/c2f/latest/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/ismir-machines/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/ismir-machines/coarse/latest/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/machines/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/machines/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/machines
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/machines/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/machines/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/machines
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/machines/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK/hugo/machines
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/machines/c2f/best/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/machines/c2f/best/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/machines/coarse/best/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/machines/coarse/best/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/musdb/c2f.yml
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/xeno-canto/c2f
|
13 |
+
|
14 |
+
AudioDataset.aligned: true
|
15 |
+
train/build_dataset.folders:
|
16 |
+
bass:
|
17 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train]
|
18 |
+
ext: "bass.wav"
|
19 |
+
drums:
|
20 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train]
|
21 |
+
ext: "drums.wav"
|
22 |
+
other:
|
23 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train]
|
24 |
+
ext: "other.wav"
|
25 |
+
vocals:
|
26 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train]
|
27 |
+
ext: "vocals.wav"
|
28 |
+
val/build_dataset.folders:
|
29 |
+
bass:
|
30 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/val]
|
31 |
+
ext: "bass.wav"
|
32 |
+
drums:
|
33 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/val]
|
34 |
+
ext: "drums.wav"
|
35 |
+
other:
|
36 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/val]
|
37 |
+
ext: "other.wav"
|
38 |
+
vocals:
|
39 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/val]
|
40 |
+
ext: "vocals.wav"
|
conf/generated/musdb/coarse.yml
ADDED
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/xeno-canto/coarse
|
6 |
+
train/build_dataset.folders:
|
7 |
+
bass:
|
8 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train/]
|
9 |
+
ext: ["bass.wav"]
|
10 |
+
drums:
|
11 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train/]
|
12 |
+
ext: ["drums.wav"]
|
13 |
+
other:
|
14 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train/]
|
15 |
+
ext: ["other.wav"]
|
16 |
+
vocals:
|
17 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/train/]
|
18 |
+
ext: ["vocals.wav"]
|
19 |
+
val/build_dataset.folders:
|
20 |
+
bass:
|
21 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/test/]
|
22 |
+
ext: ["bass.wav"]
|
23 |
+
drums:
|
24 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/test/]
|
25 |
+
ext: ["drums.wav"]
|
26 |
+
other:
|
27 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/test/]
|
28 |
+
ext: ["other.wav"]
|
29 |
+
vocals:
|
30 |
+
sources: [/media/CHONK/hugo/ial-datasets/musdb18hq/test/]
|
31 |
+
ext: ["vocals.wav"]
|
conf/generated/musdb/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Interface.coarse2fine_ckpt: ./models/vampnet/c2f.pth
|
2 |
+
Interface.coarse2fine_lora_ckpt: null
|
3 |
+
|
4 |
+
Interface.coarse_ckpt: ./runs/musdb-cond-clfdrop/best/vampnet/weights.pth
|
5 |
+
Interface.coarse_lora_ckpt: null
|
6 |
+
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/n64/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/n64/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- data/salad-bowl/n64-jungle/n64-jungle-mix.wav
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/n64/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/n64/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- data/salad-bowl/n64-jungle/n64-jungle-mix.wav
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/n64/interface.yml
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - data/salad-bowl/n64-jungle/n64-jungle-mix.wav
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/n64/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse_ckpt: ./runs/n64/coarse/latest/vampnet/weights.pth
|
5 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
6 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/natural-sounds/c2f.yml
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/soundrangers-v2/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK2/prosound_redacted/Soundrangers Complete
|
15 |
+
- /media/CHONK2/prosound_redacted/Soundrangers Update 2018
|
16 |
+
- /media/CHONK2/prosound_redacted/BBC Nature Sound Effects Library/Animals
|
17 |
+
- /media/CHONK2/prosound_redacted/BBC Nature Sound Effects Library/Birds
|
18 |
+
- /media/CHONK2/prosound_redacted/BBC Historical and 1-166 Sound Effects Library/Foley
|
19 |
+
- /media/CHONK2/prosound_redacted/BBC Historical and 1-166 Sound Effects Library/Musical
|
20 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Dogs
|
21 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Farm
|
22 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Horses
|
23 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Rodents
|
24 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Wild
|
25 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Bells
|
26 |
+
- /media/CHONK2/prosound_redacted/King Collection - Volume 1/Musical - Chimes
|
27 |
+
- /media/CHONK2/prosound_redacted/King Collection - Volume 1/Musical - Instruments
|
28 |
+
val/AudioLoader.sources: *id001
|
conf/generated/natural-sounds/coarse.yml
ADDED
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/soundrangers-v2/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK2/prosound_redacted/Soundrangers Complete
|
8 |
+
- /media/CHONK2/prosound_redacted/Soundrangers Update 2018
|
9 |
+
- /media/CHONK2/prosound_redacted/BBC Nature Sound Effects Library/Animals
|
10 |
+
- /media/CHONK2/prosound_redacted/BBC Nature Sound Effects Library/Birds
|
11 |
+
- /media/CHONK2/prosound_redacted/BBC Historical and 1-166 Sound Effects Library/Foley
|
12 |
+
- /media/CHONK2/prosound_redacted/BBC Historical and 1-166 Sound Effects Library/Musical
|
13 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Dogs
|
14 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Farm
|
15 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Horses
|
16 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Rodents
|
17 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Mammals - Wild
|
18 |
+
- /media/CHONK2/prosound_redacted/Big Room Complete/Bells
|
19 |
+
- /media/CHONK2/prosound_redacted/King Collection - Volume 1/Musical - Chimes
|
20 |
+
- /media/CHONK2/prosound_redacted/King Collection - Volume 1/Musical - Instruments
|
21 |
+
val/AudioLoader.sources: *id001
|
conf/generated/natural-sounds/interface.yml
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - /media/CHONK2/prosound_redacted/Soundrangers
|
3 |
+
- Complete
|
4 |
+
Interface.coarse2fine_ckpt: ./runs/soundrangers-v2/c2f/latest/vampnet/weights.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/soundrangers-v2/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
7 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/nes/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/nes/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- data/salad-bowl/chiptune/nes.wav
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/nes/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/nes/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- data/salad-bowl/chiptune/nes.wav
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/nes/interface.yml
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- - data/salad-bowl/chiptune/nes.wav
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/nes/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse_ckpt: ./runs/nes/coarse/latest/vampnet/weights.pth
|
5 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
6 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/nyc-subway/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/nyc-subway/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/nyc-subway
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/nyc-subway/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/nyc-subway/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/nyc-subway
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/nyc-subway/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- /media/CHONK/hugo/nyc-subway
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/nyc-subway/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/nyc-subway/c2f/latest/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/nyc-subway/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/nyc-subway/coarse/latest/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|
conf/generated/ocean-waves/c2f.yml
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
AudioDataset.duration: 3.0
|
4 |
+
AudioDataset.loudness_cutoff: -40.0
|
5 |
+
VampNet.embedding_dim: 1280
|
6 |
+
VampNet.n_codebooks: 14
|
7 |
+
VampNet.n_conditioning_codebooks: 4
|
8 |
+
VampNet.n_heads: 20
|
9 |
+
VampNet.n_layers: 16
|
10 |
+
fine_tune: true
|
11 |
+
fine_tune_checkpoint: ./models/vampnet/c2f.pth
|
12 |
+
save_path: ./runs/ocean-waves/c2f
|
13 |
+
train/AudioLoader.sources: &id001
|
14 |
+
- /media/CHONK/hugo/ocean-waves-sounds
|
15 |
+
val/AudioLoader.sources: *id001
|
conf/generated/ocean-waves/coarse.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
$include:
|
2 |
+
- conf/lora/lora.yml
|
3 |
+
fine_tune: true
|
4 |
+
fine_tune_checkpoint: ./models/vampnet/coarse.pth
|
5 |
+
save_path: ./runs/ocean-waves/coarse
|
6 |
+
train/AudioLoader.sources: &id001
|
7 |
+
- /media/CHONK/hugo/ocean-waves-sounds
|
8 |
+
val/AudioLoader.sources: *id001
|
conf/generated/ocean-waves/interface.yml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
AudioLoader.sources:
|
2 |
+
- /media/CHONK/hugo/ocean-waves-sounds.mp3
|
3 |
+
Interface.coarse2fine_ckpt: ./runs/ocean-waves/c2f/latest/vampnet/weights.pth
|
4 |
+
Interface.coarse2fine_lora_ckpt: ./runs/ocean-waves/c2f/latest/lora.pth
|
5 |
+
Interface.coarse_ckpt: ./runs/ocean-waves/coarse/latest/vampnet/weights.pth
|
6 |
+
Interface.coarse_lora_ckpt: ./runs/ocean-waves/coarse/latest/lora.pth
|
7 |
+
Interface.codec_ckpt: ./models/vampnet/codec.pth
|
8 |
+
Interface.wavebeat_ckpt: ./models/wavebeat.pth
|