surround presets

app.py

@@ -22,15 +22,15 @@ def convert_to_wav_float(input_file):
     return temp_wav.name
 
 
-def apply_reverb_wet_only(audio, samplerate):
+def apply_reverb_wet_only(audio, samplerate, reverb_args):
     """
-    Apply wet-only reverb using SoX to a single channel.
+    Apply wet-only reverb using SoX to a single channel with custom reverb args.
     """
     with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tin, \
          tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tout:
         sf.write(tin.name, audio, samplerate, subtype='FLOAT')
         subprocess.run(
-            ["sox", tin.name, tout.name, "reverb", '-w', '85', '50', '100', '95', '10', '-2'],
+            ["sox", tin.name, tout.name, "reverb", "-w"] + reverb_args,
             stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True
         )
         wet, _ = sf.read(tout.name, dtype='float32')
@@ -39,6 +39,7 @@ def apply_reverb_wet_only(audio, samplerate):
     return wet
 
 
+
 def sox_filter(audio, samplerate, filter_type, cutoff):
     """
     Apply highpass or lowpass filter via SoX.
@@ -82,38 +83,55 @@ def extract_phantom_center(input_file, rdf=0.99999):
     return fs, FL[:len(L)], FR[:len(R)], FC[:len(M)]
 
 
-def create_5_1_surround(input_file):
-    # 1. Extract fronts and centre
+def create_5_1_surround(input_file, preset="music"):
+    # Preset-based parameters
+    # Reverberance (50%) HF-damping (50%) room-scale (100%) stereo-depth (100%) pre-delay (0ms) wet-gain (0dB)
+    if preset == "music":
+        hp_cutoff = 120
+        lfe_cutoff = 120
+        reverb_args = ['85', '70', '100', '95', '10', '-2']
+    elif preset == "speech":
+        hp_cutoff = 120
+        lfe_cutoff = 120
+        reverb_args = ['50', '99', '50', '70', '0', '0']
+    elif preset == "open":
+        hp_cutoff = 120
+        lfe_cutoff = 120
+        reverb_args = ['20', '50', '100', '100', '100', '0']
+    else:
+        raise ValueError(f"Unknown preset: {preset}")
+
+    # 1. Extract FL/FR/phantom centre
     fs, FL, FR, FC = extract_phantom_center(input_file)
 
-    # 2. Read original for reverb and LFE source as float WAV
+    # 2. Get stereo original for reverb
     wav = convert_to_wav_float(input_file)
     stereo, _ = sf.read(wav, dtype='float32')
     os.unlink(wav)
-    L_orig, R_orig = stereo[:,0], stereo[:,1]
+    L_orig, R_orig = stereo[:, 0], stereo[:, 1]
 
-    # 3. Wet-only reverb for surrounds using original L/R
-    SL = apply_reverb_wet_only(L_orig, fs)
-    SR = apply_reverb_wet_only(R_orig, fs)
+    # 3. Wet-only reverb with chosen settings
+    SL = apply_reverb_wet_only(L_orig, fs, reverb_args)
+    SR = apply_reverb_wet_only(R_orig, fs, reverb_args)
 
-    # 4. High-pass filter to all but LFE
-    FL_hp = sox_filter(FL, fs, 'highpass', 120)
-    FR_hp = sox_filter(FR, fs, 'highpass', 120)
-    FC_hp = sox_filter(FC, fs, 'highpass', 120)
-    SL_hp = sox_filter(SL, fs, 'highpass', 120)
-    SR_hp = sox_filter(SR, fs, 'highpass', 120)
+    # 4. Highpass filter everything except LFE
+    FL_hp = sox_filter(FL, fs, 'highpass', hp_cutoff)
+    FR_hp = sox_filter(FR, fs, 'highpass', hp_cutoff)
+    FC_hp = sox_filter(FC, fs, 'highpass', hp_cutoff)
+    SL_hp = sox_filter(SL, fs, 'highpass', hp_cutoff)
+    SR_hp = sox_filter(SR, fs, 'highpass', hp_cutoff)
 
-    # 5. LFE from pre-filtered sum of FL/FR
+    # 5. Lowpass for LFE
     bass_sum = 0.5 * (FL + FR)
-    LFE = sox_filter(bass_sum, fs, 'lowpass', 120)
+    LFE = sox_filter(bass_sum, fs, 'lowpass', lfe_cutoff)
 
-    # 6. Pad to equal length
+    # 6. Stack and pad
     channels = [FL_hp, FR_hp, FC_hp, LFE, SL_hp, SR_hp]
     length = max(len(ch) for ch in channels)
     def pad(x): return np.pad(x, (0, length - len(x)))
     multich = np.column_stack([pad(ch) for ch in channels])
 
-    # 7. Write and encode
+    # 7. Write WAV and encode to OGG
     out_wav = tempfile.NamedTemporaryFile(suffix='.wav', delete=False)
     sf.write(out_wav.name, multich, fs, subtype='FLOAT')
     out_wav.close()
@@ -126,14 +144,22 @@ def create_5_1_surround(input_file):
     os.unlink(out_wav.name)
     return out_ogg.name
 
+
 # ========== Gradio UI ==========
 with gr.Blocks(title="Stereo to 5.1 Surround") as demo:
     gr.Markdown("# 🎧 Stereo to 5.1 OGG Converter")
-    gr.Markdown("Phantom centre extraction + wet-only reverb on original L/R + sox highpass/lowpass + float I/O")
+    gr.Markdown("Choose music or speech preset for surround processing")
+
     inp = gr.Audio(label="Upload stereo audio", type="filepath")
+    preset = gr.Dropdown(
+    label="Select Preset",
+    choices=["music", "speech", "open"],
+    value="music"  # or whichever you want as the default
+)
     btn = gr.Button("Convert to 5.1 OGG")
     out = gr.File(label="Download 5.1 OGG")
-    btn.click(fn=create_5_1_surround, inputs=[inp], outputs=[out])
+
+    btn.click(fn=create_5_1_surround, inputs=[inp, preset], outputs=[out])
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()