lots of file organization

.gitignore

@@ -165,5 +165,6 @@ cython_debug/
 data/
 .vscode/
 .DS_Store
-generated_audio.wav
+outputs/generated_audio.wav
+outputs/generated_audio_[0-9]*
 test.wav

README.md

@@ -1,17 +1,35 @@
-# StereoSampleGAN (WIP)
+# StereoSampleGAN
 
 [![On Push](https://github.com/shuklabhay/stereo-sample-gan/actions/workflows/push.yml/badge.svg)](https://github.com/shuklabhay/stereo-sample-gan/actions/workflows/push.yml/badge.svg)
 
-StereoSampleGAN: A lightweight approach high fidelity stereo audio sample generation. Generate a kick drum by running `generate.py`.
+StereoSampleGAN: A lightweight approach high fidelity stereo audio sample generation.
 
-Generated audio spectrogram examples:
-![Audio Example 1](paper/static/generated_audio_example_1.png)
-![Audio Example 2](paper/static/generated_audio_example_2.png)
-![Audio Example 3](paper/static/generated_audio_example_3.png)
+## Model Usage
+
+1. Prereqs
+
+- Optional but highly reccomended: Set up a [Python virtual environment.](https://www.youtube.com/watch?v=e5GL1obY_sI)
+  - Audio loader package `librosa` requires an outdated version of Numpy
+- Install requirements by running `pip3 install -r requirements.txt`
+
+2. Generate Audio
+
+- Specify usage paramaters in `usage_params.py`
+  - For `outputs/StereoSampleGAN-DiverseKick.pth`, `training_sample_length = 0.6`
+  - For `outputs/StereoSampleGAN-Kick.pth`, `training_sample_length = 0.6`
+- Generate audio by running `python3 generate.py`
+
+3. Train model
+
+- Specify training data paramaters in `usage_params.py`
+- Process training data by running `python3 encode_audio_data.py`
+- Train model by running `python3 stereo_sample_gan.py`
 
 ## Directories
 
-- `paper`: Research paper and static images
-- `model`: Trained model and generated audio
+- `paper`: Research paper / model writeup
+  - `static`: Static images
+- `outputs`: Trained model and generated audio
 - `src`: Model source code
   - `utils`: Model and data utilities
+  - `data_processing`: Training data processing scripts

src/data_processing/audio_processing_validation.py

@@ -5,7 +5,7 @@
 
 import os
 import random
-from usage.usage_specs import compiled_data_path
+from usage_params import compiled_data_path
 from utils.signal_helpers import (
     stft_and_istft,
 )

src/data_processing/encode_audio_data.py

@@ -8,7 +8,7 @@
     load_loudness_data,
 )
 from utils.signal_helpers import encode_sample_directory
-from usage.usage_specs import training_audio_dir, compiled_data_path
+from usage_params import training_audio_dir, compiled_data_path
 
 # Encode audio samples
 if len(sys.argv) > 1:

src/usage/generate.py → src/generate.py

@@ -1,12 +1,7 @@
-import sys
 import os
-
-sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-
-
 import torch
 from architecture import Generator, LATENT_DIM
-from usage.usage_specs import (
+from usage_params import (
     model_to_generate_with,
     outputs_dir,
     generated_audio_name,
@@ -41,7 +36,7 @@
     current_sample = generated_output[i]
 
     audio_info = norm_db_to_audio(current_sample)
-    audio_save_path = os.path.join(outputs_dir, f"{generated_audio_name}-{i + 1}.wav")
+    audio_save_path = os.path.join(outputs_dir, f"{generated_audio_name}_{i + 1}.wav")
 
     save_audio(audio_save_path, audio_info)
 

src/stereo_sample_gan.py

@@ -1,9 +1,3 @@
-import sys
-import os
-
-sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-
-
 import torch
 from torch.utils.data import DataLoader, TensorDataset, random_split
 
@@ -16,7 +10,7 @@
     load_loudness_data,
 )
 
-from usage.usage_specs import compiled_data_path
+from usage_params import compiled_data_path
 
 # Constants
 LR_G = 0.003

src/usage/usage_specs.py → src/usage_params.py

@@ -1,16 +1,18 @@
 # Processing training data
-training_audio_dir = "data/one_shots"  # Your training data path
+training_audio_dir = "data/kick_samples_diverse"  # Your training data path
 compiled_data_path = "data/compiled_data.npy"  # Your compiled data/output path
 training_sample_length = 0.6  # seconds
 
 # Saving model
 outputs_dir = "outputs"  # Where to save your generated audio & model
-model_save_name = "StereoSampleGAN-OldKick"  # What to name your model save
+model_save_name = "StereoSampleGAN-DiverseKick"  # What to name your model save
 model_save_path = f"{outputs_dir}/{model_save_name}.pth"
 
 # Generating audio
 model_to_generate_with = model_save_path  # Generation model path
 audio_generation_count = 2  # Audio examples to generate
 generated_audio_name = "generated_audio"  # Output file name
-generated_sample_length = 0.6  # Match model training data audio length
-visualize_generated = True  # SHow generated audio spectrogra,s
+generated_sample_length = (
+    training_sample_length  # Match model training data audio length
+)
+visualize_generated = True  # Show generated audio spectrograms

src/utils/file_helpers.py

@@ -3,7 +3,7 @@
 import torch
 import soundfile as sf
 
-from usage.usage_specs import model_save_path
+from usage_params import model_save_path
 
 # Constants
 GLOBAL_SR = 44100

src/utils/signal_helpers.py

@@ -10,7 +10,7 @@
 import plotly.subplots as sp
 import scipy
 
-from usage.usage_specs import (
+from usage_params import (
     training_sample_length,
     outputs_dir,
 )