Upload

app.py

@@ -58,15 +58,15 @@ model.eval()
 
    
 @spaces.GPU
-def enhance(filepath):
-    wav_np, orig_sr = librosa.load(filepath, sr=None)
-    if orig_sr!=SR: wav_np=librosa.resample(wav_np,orig_sr,SR)
+def enhance(path):
+    wav_np, orig_sr = librosa.load(path, sr=None)
+    if orig_sr!=SR: wav_np=librosa.resample(wav_np, orig_sr, SR)
     wav = torch.from_numpy(wav_np).float().to(device)
     norm= torch.sqrt(len(wav)/torch.sum(wav**2)); wav=(wav*norm).unsqueeze(0)
     amp,pha,_ = mag_phase_stft(wav,**stft_cfg,compress_factor=model_cfg["compress_factor"])
     amp_g,pha_g = model(amp,pha)
     out = mag_phase_istft(amp_g,pha_g,**stft_cfg,compress_factor=model_cfg["compress_factor"])
-    out = (out/norm).squeeze().cpu().numpy()
+    out=(out/norm).squeeze().cpu().numpy()
     if orig_sr!=SR: out=librosa.resample(out,SR,orig_sr)
     sf.write("enhanced.wav", out, orig_sr)
 
@@ -74,35 +74,44 @@ def enhance(filepath):
     S = librosa.amplitude_to_db(np.abs(D), ref=np.max)
     fig,ax=plt.subplots(figsize=(6,3))
     librosa.display.specshow(S,sr=orig_sr,hop_length=512,x_axis="time",y_axis="hz",ax=ax)
-    ax.set_title("Enhanced Spectrogram"); plt.colorbar(format="%+2.0f dB",ax=ax)
+    ax.set_title("Enhanced Spectrogram"); plt.colorbar(format="%+2.0f dB",ax=ax)
     return "enhanced.wav", fig
 
-# — Custom CSS —
+# — custom CSS to remove all borders and style panels —
 CSS = """
-#title {text-align:center; margin-bottom:0.2em;}
-#subtitle {text-align:center; color:#555; margin-bottom:1.5em;}
-.duplicate-button {display:block; margin:0 auto 1.5em;}
-#audio-in {border:2px dashed #aaa; border-radius:8px; padding:1em;}
-#run-btn {width:100%; margin-top:0.5em;}
-#out-audio, #spec-plot {margin-top:1em;}
+.gr-block {padding:2rem;}
+#input-audio .gradio-component, #output-audio .gradio-component, #spec-plot .gradio-component {
+  border:1px solid #ddd; border-radius:8px; background:#fafafa; padding:1rem;
+}
+#input-audio label, #output-audio label, #spec-plot label {font-weight:600; color:#333;}
+#run-btn {width:100%; margin-top:1rem;}
 """
 
-# — Blocks layout —
-with gr.Blocks(css=CSS, theme="soft") as demo:
-    gr.HTML("<h1 id='title'>🎧 SEMamba: Speech Enhancement</h1>")
-    gr.HTML("<p id='subtitle'>Upload or record your noisy clip, then click Enhance to boost clarity and view its spectrogram.</p>")
-    gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
+with gr.Blocks(css=CSS, theme=gr.themes.Default()) as demo:
+    gr.HTML("<h2 style='text-align:center;'>🎧 SEMamba: Speech Enhancement</h2>")
+    gr.HTML("<p style='text-align:center;color:#555;'>Upload or record noisy speech, then click Enhance to clean it.</p>")
 
     with gr.Row():
-        with gr.Column(scale=1):
-            audio_in  = gr.Audio(sources=["upload","microphone"], type="filepath",
-                                 label="Your Noisy Audio", elem_id="audio-in")
-            run_btn   = gr.Button("Enhance Now 🚀", variant="primary", elem_id="run-btn")
-
-        with gr.Column(scale=1):
-            audio_out = gr.Audio(type="filepath", label="Enhanced Audio", elem_id="out-audio")
-            spec_plot = gr.Plot(label="Spectrogram", elem_id="spec-plot")
-
-    run_btn.click(enhance, inputs=audio_in, outputs=[audio_out, spec_plot])
+        with gr.Column():
+            input_audio = gr.Audio(
+                sources=["upload","microphone"],
+                type="filepath",
+                label="Input Audio",
+                elem_id="input-audio"
+            )
+            run_btn = gr.Button("Enhance", variant="primary", elem_id="run-btn")
+
+        with gr.Column():
+            output_audio = gr.Audio(
+                type="filepath",
+                label="Enhanced Audio",
+                elem_id="output-audio"
+            )
+            spec_plot = gr.Plot(
+                label="Enhanced Spectrogram",
+                elem_id="spec-plot"
+            )
+
+    run_btn.click(enhance, inputs=input_audio, outputs=[output_audio, spec_plot])
 
 demo.launch()