Update app.py

app.py

@@ -6,17 +6,20 @@ import os
 import noisereduce as nr
 from scipy.io import wavfile
 import subprocess
-import shutil
+import torch
+from demucs import pretrained
+from demucs.apply import apply_model
+from demucs.audio import load_audio, save_audio
 
 # Helper functions
 def audiosegment_to_array(audio):
     return np.array(audio.get_array_of_samples()), audio.frame_rate
 
-def array_to_audiosegment(samples, frame_rate, sample_width=2, channels=1):
+def array_to_audiosegment(samples, frame_rate, channels=1):
     return AudioSegment(
         samples.tobytes(),
         frame_rate=frame_rate,
-        sample_width=sample_width,
+        sample_width=samples.dtype.itemsize,
         channels=channels
     )
 
@@ -66,68 +69,76 @@ def apply_bass_boost(audio, gain=10):
 def apply_treble_boost(audio, gain=10):
     return audio.high_pass_filter(4000).apply_gain(gain)
 
-# Save temp WAV file for spleeter
-def save_temp_wav(audio, path):
-    audio.export(path, format="wav")
-
-# Main processing function
-def process_audio(audio_file, effect):
+# Vocal Isolation using Demucs
+def apply_vocal_isolation(audio_path):
+    model = pretrained.get_model(name='htdemucs')
+    wav = load_audio(audio_path)
+    ref = wav.mean(0)
+    wav -= ref[:, None]
+    sources = apply_model(model, wav[None])[0]
+    wav += ref[:, None]
+    vocal_track = sources[3]  # index 3 = vocals
+    out_path = os.path.join(tempfile.gettempdir(), "vocals.wav")
+    save_audio(vocal_track, out_path, samplerate=model.samplerate)
+    return out_path
+
+# Apply selected effects in order
+def process_audio(audio_file, effects, isolate_vocals):
     audio = AudioSegment.from_file(audio_file)
-
-    if effect == "Normalize":
-        result = apply_normalize(audio)
-    elif effect == "Noise Reduction":
-        result = apply_noise_reduction(audio)
-    elif effect == "Compress Dynamic Range":
-        result = apply_compression(audio)
-    elif effect == "Add Reverb":
-        result = apply_reverb(audio)
-    elif effect == "Pitch Shift":
-        result = apply_pitch_shift(audio)
-    elif effect == "Echo":
-        result = apply_echo(audio)
-    elif effect == "Stereo Widening":
-        result = apply_stereo_widen(audio)
-    elif effect == "Bass Boost":
-        result = apply_bass_boost(audio)
-    elif effect == "Treble Boost":
-        result = apply_treble_boost(audio)
-    elif effect == "Isolate Vocals":
-        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as f:
-            save_temp_wav(audio, f.name)
-            vocal_path = apply_vocal_isolation(f.name)
-            result = AudioSegment.from_wav(vocal_path)
-    else:
-        result = audio
+    original = audio
+
+    effect_map = {
+        "Noise Reduction": apply_noise_reduction,
+        "Compress Dynamic Range": apply_compression,
+        "Add Reverb": apply_reverb,
+        "Pitch Shift": lambda x: apply_pitch_shift(x),
+        "Echo": apply_echo,
+        "Stereo Widening": apply_stereo_widen,
+        "Bass Boost": apply_bass_boost,
+        "Treble Boost": apply_treble_boost,
+        "Normalize": apply_normalize,
+    }
+
+    for effect_name in effects:
+        if effect_name in effect_map:
+            audio = effect_map[effect_name](audio)
 
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
-        result.export(f.name, format="wav")
+        if isolate_vocals:
+            temp_input = os.path.join(tempfile.gettempdir(), "input.wav")
+            audio.export(temp_input, format="wav")
+            vocal_path = apply_vocal_isolation(temp_input)
+            final_audio = AudioSegment.from_wav(vocal_path)
+        else:
+            final_audio = audio
+
+        final_audio.export(f.name, format="wav")
         return f.name
 
 # Gradio Interface
+effect_choices = [
+    "Noise Reduction",
+    "Compress Dynamic Range",
+    "Add Reverb",
+    "Pitch Shift",
+    "Echo",
+    "Stereo Widening",
+    "Bass Boost",
+    "Treble Boost",
+    "Normalize"
+]
+
 interface = gr.Interface(
     fn=process_audio,
     inputs=[
         gr.Audio(label="Upload Audio", type="filepath"),
-        gr.Dropdown(
-            choices=[
-                "Normalize",
-                "Noise Reduction",
-                "Compress Dynamic Range",
-                "Add Reverb",
-                "Pitch Shift",
-                "Echo",
-                "Stereo Widening",
-                "Bass Boost",
-                "Treble Boost",
-               
-            ],
-            label="Select Effect"
-        )
+        gr.CheckboxGroup(choices=effect_choices, label="Apply Effects in Order"),
+        gr.Checkbox(label="Isolate Vocals After Effects")
     ],
     outputs=gr.Audio(label="Processed Audio", type="filepath"),
-    title="Fix My Recording - Pro Edition",
-    description="Apply noise reduction, mastering, creative effects, or isolate vocals from your track!"
+    title="Fix My Recording - Studio Pro",
+    description="Apply multiple effects in sequence and optionally isolate vocals!",
+    allow_flagging="never"
 )
 
 interface.launch()