add colab

.gitignore

@@ -3,3 +3,4 @@ __pycache__
 .ipynb_checkpoints
 data*
 ddpm-ema-audio-*
+flagged

README.md

@@ -22,7 +22,7 @@ Audio can be represented as images by transforming to a [mel spectrogram](https:
 
 A DDPM model is trained on a set of mel spectrograms that have been generated from a directory of audio files. It is then used to synthesize similar mel spectrograms, which are then converted back into audio. See the `test-model.ipynb` notebook for an example.
 
-You can play around with the model I trained on about 500 songs from my Spotify "liked" playlist [here](https://huggingface.co/spaces/teticio/audio-diffusion)
+You can play around with the model I trained on about 500 songs from my Spotify "liked" playlist on [Google Colab](https://colab.research.google.com/github/teticio/audio-diffusion/blob/master/notebooks/test-model.ipynb) or [Hugging Face spaces](https://huggingface.co/spaces/teticio/audio-diffusion).
 
 ## Generate Mel spectrogram dataset from directory of audio files
 #### Training can be run with Mel spectrograms of resolution 64x64 on a single commercial grade GPU (e.g. RTX 2080 Ti). The `hop_length` should be set to 1024 for better results.

app.py

@@ -28,7 +28,8 @@ if __name__ == "__main__":
     demo = gr.Interface(
         fn=generate_spectrogram_and_audio,
         title="Audio Diffusion",
-        description=f"Generate audio using Huggingface diffusers",
+        description=f"Generate audio using Huggingface diffusers.\
+            This takes about 20 minutes without a GPU, so why not make yourself a cup of tea in the meantime?",
         inputs=[],
         outputs=[
             gr.Image(label="Mel spectrogram", image_mode="L"),

notebooks/test-model.ipynb

@@ -1,5 +1,30 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "0fd939b0",
+   "metadata": {},
+   "source": [
+    "<a href=\"https://colab.research.google.com/github/teticio/audio-diffusion/blob/master/notebooks/test-model.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "6c7800a6",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "try:\n",
+    "    # are we running on Google Colab?\n",
+    "    import google.colab\n",
+    "    !git clone -q https://github.com/teticio/audio-diffusion.git\n",
+    "    %cd audio-diffusion\n",
+    "    !pip install -q -r requirements.txt\n",
+    "except:\n",
+    "    pass"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 1,
@@ -256,22 +281,29 @@
     "\n",
     "\n",
     "args = AttributeDict({\n",
-    "    \"hub_model_id\": \"teticio/audio-diffusion-256\",\n",
-    "    \"output_dir\": \"../ddpm-ema-audio-256-repo\",\n",
-    "    \"local_rank\": -1,\n",
-    "    \"hub_token\": \"hf_\",\n",
-    "    \"hub_private_repo\": False,\n",
-    "    \"overwrite_output_dir\": False\n",
+    "    \"hub_model_id\":\n",
+    "    \"teticio/audio-diffusion-256\",\n",
+    "    \"output_dir\":\n",
+    "    \"../ddpm-ema-audio-256-repo\",\n",
+    "    \"local_rank\":\n",
+    "    -1,\n",
+    "    \"hub_token\":\n",
+    "    open(os.path.join(os.environ['HOME'], '.huggingface/token'), 'rt').read(),\n",
+    "    \"hub_private_repo\":\n",
+    "    False,\n",
+    "    \"overwrite_output_dir\":\n",
+    "    False\n",
     "})\n",
     "\n",
     "repo = init_git_repo(args, at_init=True)\n",
+    "ddpm = DDPMPipeline.from_pretrained('../ddpm-ema-audio-256')\n",
     "push_to_hub(args, ddpm, repo)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "11b3963b",
+   "id": "8c8261a0",
    "metadata": {},
    "outputs": [],
    "source": []

requirements-lock.txt

@@ -11,7 +11,7 @@ argon2-cffi==21.3.0
 argon2-cffi-bindings==21.2.0
 async-timeout==4.0.2
 attrs==21.4.0
-audioread==2.1.9
+audioread==3.0.0
 backcall==0.2.0
 backoff==1.10.0
 bcrypt==3.2.2
@@ -25,9 +25,12 @@ cachetools==5.2.0
 captum==0.5.0
 certifi==2022.6.15
 cffi==1.15.1
+cfgv==3.3.1
 charset-normalizer==2.1.0
 click==8.1.3
 cloudpickle==2.1.0
+colossalai==0.1.8
+commonmark==0.9.1
 cryptography==37.0.4
 cycler==0.11.0
 datasets==2.4.0
@@ -35,9 +38,11 @@ debugpy==1.6.2
 decorator==5.1.1
 deepspeed==0.7.0
 defusedxml==0.7.1
-diffusers==0.1.3
+diffusers==0.2.2
 dill==0.3.5.1
+distlib==0.3.5
 entrypoints==0.4
+fabric==2.7.1
 fastapi==0.79.0
 fastjsonschema==2.16.1
 ffmpy==0.3.0
@@ -51,12 +56,14 @@ google-pasta==0.2.0
 gradio==3.1.4
 grpcio==1.47.0
 h11==0.12.0
-hjson==3.0.2
+hjson==3.1.0
 httpcore==0.15.0
 httpx==0.23.0
 huggingface-hub==0.8.1
+identify==2.5.3
 idna==3.3
 importlib-metadata==4.12.0
+invoke==1.7.1
 ipykernel==6.15.1
 ipython==7.34.0
 ipython-genutils==0.2.0
@@ -78,10 +85,10 @@ lxml==4.9.1
 Markdown==3.4.1
 markdown-it-py==2.1.0
 MarkupSafe==2.1.1
-matplotlib==3.5.2
+matplotlib==3.5.3
 matplotlib-inline==0.1.3
 mdit-py-plugins==0.3.0
-mdurl==0.1.1
+mdurl==0.1.2
 mistune==0.8.4
 monotonic==1.6
 more-itertools==8.14.0
@@ -90,23 +97,25 @@ multiprocess==0.70.13
 munkres==1.1.4
 mypy-extensions==0.4.3
 nbclient==0.6.6
-nbconvert==6.5.1
+nbconvert==6.5.3
 nbformat==5.4.0
 nest-asyncio==1.5.5
 networkx==2.8.5
 ninja==1.10.2.3
 nlp==0.4.0
 nltk==3.7
+nodeenv==1.7.0
 notebook==6.4.12
 numba==0.56.0
 numpy==1.22.4
 oauthlib==3.2.0
-orjson==3.7.11
+orjson==3.7.12
 packaging==21.3
 pandas==1.4.3
 pandocfilters==1.5.0
 paramiko==2.11.0
 parso==0.8.3
+pathlib2==2.3.7.post1
 pathos==0.2.9
 pathspec==0.9.0
 pexpect==4.8.0
@@ -117,6 +126,7 @@ pluggy==0.13.1
 pooch==1.6.0
 pox==0.3.1
 ppft==1.7.6.5
+pre-commit==2.20.0
 prometheus-client==0.14.1
 prompt-toolkit==3.0.30
 protobuf==3.19.4
@@ -130,7 +140,7 @@ pyasn1==0.4.8
 pyasn1-modules==0.2.8
 pycparser==2.21
 pycryptodome==3.15.0
-pydantic==1.9.1
+pydantic==1.9.2
 pydub==0.25.1
 Pygments==2.12.0
 PyNaCl==1.5.0
@@ -140,15 +150,16 @@ pytest==5.4.3
 python-dateutil==2.8.2
 python-dotenv==0.20.0
 python-multipart==0.0.5
-pytz==2022.1
+pytz==2022.2.1
 PyYAML==6.0
-pyzmq==23.2.0
+pyzmq==23.2.1
 regex==2022.7.25
 requests==2.28.1
 requests-oauthlib==1.3.1
 resampy==0.4.0
 responses==0.18.0
 rfc3986==1.5.0
+rich==12.5.1
 rsa==4.9
 s3fs==2022.7.1
 s3transfer==0.5.2
@@ -167,7 +178,7 @@ snorkel==0.9.9
 SoundFile==0.10.3.post1
 soupsieve==2.3.2.post1
 starlette==0.19.1
-tensorboard==2.9.1
+tensorboard==2.10.0
 tensorboard-data-server==0.6.1
 tensorboard-plugin-wit==1.8.1
 terminado==0.15.0
@@ -176,8 +187,9 @@ tinycss2==1.1.1
 tokenizers==0.12.1
 toml==0.10.2
 tomli==2.0.1
-torch==1.12.1
-torchvision==0.13.1
+torch==1.12.1+cu116
+torchaudio==0.12.1+cu116
+torchvision==0.13.1+cu116
 tornado==6.2
 tqdm==4.64.0
 traitlets==5.3.0
@@ -187,6 +199,7 @@ typing_extensions==4.3.0
 uc-micro-py==1.0.1
 urllib3==1.26.11
 uvicorn==0.18.2
+virtualenv==20.16.3
 wcwidth==0.2.5
 webencodings==0.5.1
 Werkzeug==2.2.2

requirements.txt

@@ -1,6 +1,7 @@
-# for Hugging Face spaces
+# for Hugging Face Spaces
 torch
 numpy
 Pillow
 diffusers
 librosa
+datasets

src/train_unconditional.py

@@ -241,13 +241,16 @@ def main(args):
             if epoch % args.save_model_epochs == 0 or epoch == args.num_epochs - 1:
                 # save the model
                 if args.push_to_hub:
-                    push_to_hub(
-                        args,
-                        pipeline,
-                        repo,
-                        commit_message=f"Epoch {epoch}",
-                        blocking=False,
-                    )
+                    try:
+                        push_to_hub(
+                            args,
+                            pipeline,
+                            repo,
+                            commit_message=f"Epoch {epoch}",
+                            blocking=False,
+                        )
+                    except NameError:  # current version of diffusers has a little bug
+                        pass
                 else:
                     pipeline.save_pretrained(output_dir)
         accelerator.wait_for_everyone()