Spaces:
Runtime error
Runtime error
add examples
Browse files
app.py
CHANGED
|
@@ -173,22 +173,26 @@ with gr.Blocks() as demo:
|
|
| 173 |
gr.Markdown("Interactive demo of Inference Time Optimization (ITO) for Music Mastering Style Transfer. \
|
| 174 |
The mastering style transfer is performed by a differentiable audio processing model, and the predicted parameters are shown as the output. \
|
| 175 |
Perform mastering style transfer with an input source audio and a reference mastering style audio. On top of this result, you can perform ITO to optimize the reference embedding $z_{ref}$ to further gain control over the output mastering style.")
|
| 176 |
-
gr.Image("ito_snow.png", width=
|
| 177 |
|
| 178 |
gr.Markdown("## Step 1: Mastering Style Transfer")
|
| 179 |
|
| 180 |
with gr.Tab("Upload Audio"):
|
| 181 |
with gr.Row():
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
|
|
|
|
|
|
|
|
|
| 185 |
# Dropdowns for selecting example files
|
| 186 |
with gr.Row():
|
| 187 |
-
input_example_dropdown = gr.Dropdown(label="Select Input Example", choices=input_examples)
|
| 188 |
-
reference_example_dropdown = gr.Dropdown(label="Select Reference Example", choices=reference_examples)
|
| 189 |
|
| 190 |
process_button = gr.Button("Process Mastering Style Transfer")
|
| 191 |
gr.Markdown('<span style="color: lightgray; font-style: italic;">all output samples are normalized to -12dB</span>')
|
|
|
|
| 192 |
|
| 193 |
with gr.Row():
|
| 194 |
with gr.Column():
|
|
@@ -196,19 +200,27 @@ with gr.Blocks() as demo:
|
|
| 196 |
normalized_input = gr.Audio(label="Normalized Source Audio", type='numpy')
|
| 197 |
param_output = gr.Textbox(label="Predicted Parameters", lines=5)
|
| 198 |
|
| 199 |
-
def process_audio_with_examples(input_audio, reference_audio, input_example, reference_example):
|
| 200 |
-
if
|
| 201 |
-
|
| 202 |
-
|
| 203 |
-
|
|
|
|
| 204 |
return process_audio(input_audio, reference_audio)
|
| 205 |
|
| 206 |
process_button.click(
|
| 207 |
process_audio_with_examples,
|
| 208 |
-
inputs=[input_audio, reference_audio, input_example_dropdown, reference_example_dropdown],
|
| 209 |
outputs=[output_audio, param_output, normalized_input]
|
| 210 |
)
|
| 211 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 212 |
with gr.Tab("YouTube Audio"):
|
| 213 |
with gr.Row():
|
| 214 |
input_youtube_url = gr.Textbox(label="Input YouTube URL")
|
|
|
|
| 173 |
gr.Markdown("Interactive demo of Inference Time Optimization (ITO) for Music Mastering Style Transfer. \
|
| 174 |
The mastering style transfer is performed by a differentiable audio processing model, and the predicted parameters are shown as the output. \
|
| 175 |
Perform mastering style transfer with an input source audio and a reference mastering style audio. On top of this result, you can perform ITO to optimize the reference embedding $z_{ref}$ to further gain control over the output mastering style.")
|
| 176 |
+
gr.Image("ito_snow.png", width=300, height=200, label="ITO pipeline")
|
| 177 |
|
| 178 |
gr.Markdown("## Step 1: Mastering Style Transfer")
|
| 179 |
|
| 180 |
with gr.Tab("Upload Audio"):
|
| 181 |
with gr.Row():
|
| 182 |
+
use_example_files = gr.Checkbox(label="Use Example Files", value=False)
|
| 183 |
+
|
| 184 |
+
with gr.Row():
|
| 185 |
+
input_audio = gr.Audio(label="Source Audio $x_{in}$", interactive=not use_example_files.value)
|
| 186 |
+
reference_audio = gr.Audio(label="Reference Style Audio $x_{ref}$", interactive=not use_example_files.value)
|
| 187 |
+
|
| 188 |
# Dropdowns for selecting example files
|
| 189 |
with gr.Row():
|
| 190 |
+
input_example_dropdown = gr.Dropdown(label="Select Input Example", choices=input_examples, interactive=use_example_files.value)
|
| 191 |
+
reference_example_dropdown = gr.Dropdown(label="Select Reference Example", choices=reference_examples, interactive=use_example_files.value)
|
| 192 |
|
| 193 |
process_button = gr.Button("Process Mastering Style Transfer")
|
| 194 |
gr.Markdown('<span style="color: lightgray; font-style: italic;">all output samples are normalized to -12dB</span>')
|
| 195 |
+
gr.Markdown("all output samples are normalized to -12dB")
|
| 196 |
|
| 197 |
with gr.Row():
|
| 198 |
with gr.Column():
|
|
|
|
| 200 |
normalized_input = gr.Audio(label="Normalized Source Audio", type='numpy')
|
| 201 |
param_output = gr.Textbox(label="Predicted Parameters", lines=5)
|
| 202 |
|
| 203 |
+
def process_audio_with_examples(input_audio, reference_audio, input_example, reference_example, use_examples):
|
| 204 |
+
if use_examples:
|
| 205 |
+
if input_example:
|
| 206 |
+
input_audio = sf.read(os.path.join(EXAMPLES_DIR, input_example))[0]
|
| 207 |
+
if reference_example:
|
| 208 |
+
reference_audio = sf.read(os.path.join(EXAMPLES_DIR, reference_example))[0]
|
| 209 |
return process_audio(input_audio, reference_audio)
|
| 210 |
|
| 211 |
process_button.click(
|
| 212 |
process_audio_with_examples,
|
| 213 |
+
inputs=[input_audio, reference_audio, input_example_dropdown, reference_example_dropdown, use_example_files],
|
| 214 |
outputs=[output_audio, param_output, normalized_input]
|
| 215 |
)
|
| 216 |
|
| 217 |
+
# Update the interactivity of the audio inputs and dropdowns based on the checkbox
|
| 218 |
+
use_example_files.change(
|
| 219 |
+
lambda use_examples: (not use_examples, not use_examples),
|
| 220 |
+
inputs=[use_example_files],
|
| 221 |
+
outputs=[input_audio, reference_audio]
|
| 222 |
+
)
|
| 223 |
+
|
| 224 |
with gr.Tab("YouTube Audio"):
|
| 225 |
with gr.Row():
|
| 226 |
input_youtube_url = gr.Textbox(label="Input YouTube URL")
|