setup to train curatedkick

.gitignore

@@ -165,7 +165,7 @@ cython_debug/
 data/
 .vscode/
 .DS_Store
-paper/training_progress
+outputs/training_progress
 outputs/generated_audio.wav
 outputs/generated_audio_[0-9]*
 test.wav

README.md

@@ -30,7 +30,7 @@ StereoSampleGAN: A lightweight approach high fidelity stereo audio sample genera
 
 ### Diverse Kick Drums
 
-Kick Drum generation model trained on ~8000 essentially random kick drums.
+Kick drum generation model trained on ~8000 essentially random kick drums.
 
 - More variation between each generated sample but audio is often inconsistent and contains some artifacts.
 - `model_save_name="StereoSampleGAN-DiverseKick`
@@ -42,7 +42,7 @@ Training progress:
 
 ### Diverse Kick Drums
 
-Kick Drum generation model trained on ~4000 curated kick drums.
+Kick drum generation model trained on ~4400 curated kick drums.
 
 - Less variation between each drum sample but also less noisy and closer to the "normal" kick drum sound
 - `model_save_name="StereoSampleGAN-CuratedKick`

src/usage_params.py

@@ -1,13 +1,12 @@
 # Main params
 training_sample_length = 0.6  # seconds
-model_save_name = "StereoSampleGAN-DiverseKick"  # What to name your model save
+outputs_dir = "outputs"  # Where to save your generated audio & model
 
-# Processing training data
-training_audio_dir = "data/kick_samples_diverse"  # Your training data path
-compiled_data_path = "data/compiled_data.npy"  # Your compiled data/output path
+model_save_name = "StereoSampleGAN-CuratedKick"  # What to name your model save
+training_audio_dir = "data/kick_samples_curated"  # Your training data path
 
-# Saving model
-outputs_dir = "outputs"  # Where to save your generated audio & model
+# Training
+compiled_data_path = "data/compiled_data.npy"  # Your compiled data/output path
 model_save_path = f"{outputs_dir}/{model_save_name}.pth"
 
 # Generating audio

src/utils/signal_helpers.py

@@ -194,7 +194,7 @@ def graph_spectrogram(audio_data, sample_name, save=False):
     if save is False:
         fig.show()
     else:
-        fig.write_image(f"paper/training_progress/{sample_name}")
+        fig.write_image(f"outputs/training_progress/{sample_name}")
 
 
 def generate_sine_impulses(num_impulses=1, outPath="model"):