Update app.py

app.py

@@ -138,7 +138,7 @@ def match_loudness(audio_path, target_lufs=-14.0):
     adjusted.export(out_path, format="wav")
     return out_path
 
-# === AI Mastering Chain – Genre EQ + Loudness Match ===
+# === AI Mastering Chain – Genre EQ + Loudness Match + Limiting ===
 def ai_mastering_chain(audio_path, genre="Pop", target_lufs=-14.0):
     audio = AudioSegment.from_file(audio_path)
 
@@ -154,6 +154,9 @@ def ai_mastering_chain(audio_path, genre="Pop", target_lufs=-14.0):
     gain_db = target_lufs - loudness
     final_audio = eq_audio + gain_db
 
+    # Apply final limiting
+    final_audio = apply_limiter(final_audio)
+
     out_path = os.path.join(tempfile.gettempdir(), "mastered_output.wav")
     final_audio.export(out_path, format="wav")
     return out_path
@@ -188,90 +191,64 @@ def auto_eq(audio, genre="Pop"):
 
     return array_to_audiosegment(samples.astype(np.int16), sr, channels=audio.channels)
 
-# === Multiband Compression ===
-def multiband_compression(audio, low_gain=0, mid_gain=0, high_gain=0):
-    samples, sr = audiosegment_to_array(audio)
-    samples = samples.astype(np.float64)
-
-    # Low Band: 20–500Hz
-    sos_low = butter(10, [20, 500], btype='band', output='sos', fs=sr)
-    low_band = sosfilt(sos_low, samples)
-    low_compressed = low_band * (10 ** (low_gain / 20))
-
-    # Mid Band: 500–4000Hz
-    sos_mid = butter(10, [500, 4000], btype='band', output='sos', fs=sr)
-    mid_band = sosfilt(sos_mid, samples)
-    mid_compressed = mid_band * (10 ** (mid_gain / 20))
-
-    # High Band: 4000–20000Hz
-    sos_high = butter(10, [4000, 20000], btype='high', output='sos', fs=sr)
-    high_band = sosfilt(sos_high, samples)
-    high_compressed = high_band * (10 ** (high_gain / 20))
-
-    total = low_compressed + mid_compressed + high_compressed
-    return array_to_audiosegment(total.astype(np.int16), sr, channels=audio.channels)
-
-# === Real-Time Spectrum Analyzer + EQ Preview ===
-def visualize_spectrum(audio_path):
-    y, sr = torchaudio.load(audio_path)
-    y_np = y.numpy().flatten()
-    
-    stft = librosa.stft(y_np)
-    db = librosa.amplitude_to_db(abs(stft))
-
-    plt.figure(figsize=(10, 4))
-    img = librosa.display.specshow(db, sr=sr, x_axis="time", y_axis="hz", cmap="magma")
-    plt.colorbar(img, format="%+2.0f dB")
-    plt.title("Frequency Spectrum")
-    plt.tight_layout()
-    buf = BytesIO()
-    plt.savefig(buf, format="png")
-    plt.close()
-    buf.seek(0)
-    return Image.open(buf)
-
-# === Stereo Imaging Tool ===
-def stereo_imaging(audio, mid_side_balance=0.5, stereo_wide=1.0):
-    mid = audio.pan(0)
-    side = audio.pan(0.3)
-    return audio.overlay(side, position=0)
-
-# === Harmonic Saturation ===
-def harmonic_saturation(audio, intensity=0.2):
+# === Harmonic Saturation / Exciter ===
+def harmonic_saturation(audio, saturation_type="Tube", intensity=0.2):
     samples = np.array(audio.get_array_of_samples()).astype(np.float32)
-    distorted = np.tanh(intensity * samples)
-    return array_to_audiosegment(distorted.astype(np.int16), audio.frame_rate, channels=audio.channels)
 
-# === Sidechain Compression ===
-def sidechain_compressor(main, sidechain, threshold=-16, ratio=4, attack=5, release=200):
-    main_seg = AudioSegment.from_file(main)
-    sidechain_seg = AudioSegment.from_file(sidechain)
-    return main_seg.overlay(sidechain_seg - 10)
+    if saturation_type == "Tube":
+        saturated = np.tanh(intensity * samples)
+    elif saturation_type == "Tape":
+        saturated = np.where(samples > 0, 1 - np.exp(-intensity * samples), -1 + np.exp(intensity * samples))
+    elif saturation_type == "Console":
+        saturated = np.clip(samples, -32768, 32768) * intensity
+    elif saturation_type == "Mix Bus":
+        saturated = np.log1p(np.abs(samples)) * np.sign(samples) * intensity
+    else:
+        saturated = samples
+
+    return array_to_audiosegment(saturated.astype(np.int16), audio.frame_rate, channels=audio.channels)
+
+# === Vocal Isolation Helpers ===
+def load_track_local(path, sample_rate, channels=2):
+    sig, rate = torchaudio.load(path)
+    if rate != sample_rate:
+        sig = torchaudio.functional.resample(sig, rate, sample_rate)
+    if channels == 1:
+        sig = sig.mean(0)
+    return sig
+
+def save_track(path, wav, sample_rate):
+    path = Path(path)
+    torchaudio.save(str(path), wav, sample_rate)
+
+def apply_vocal_isolation(audio_path):
+    model = pretrained.get_model(name='htdemucs')
+    wav = load_track_local(audio_path, model.samplerate, channels=2)
+    ref = wav.mean(0)
+    wav -= ref[:, None]
+    sources = apply_model(model, wav[None])[0]
+    wav += ref[:, None]
+
+    vocal_track = sources[3].cpu()
+    out_path = os.path.join(tempfile.gettempdir(), "vocals.wav")
+    save_track(out_path, vocal_track, model.samplerate)
+    return out_path
 
-# === Vocal Pitch Correction – Auto-Tune Style ===
-def auto_tune_vocal(audio, target_key="C"):
-    return apply_pitch_shift(audio, 0.2)
+# === Stem Splitting (Drums, Bass, Other, Vocals) ===
+def stem_split(audio_path):
+    model = pretrained.get_model(name='htdemucs')
+    wav = load_track_local(audio_path, model.samplerate, channels=2)
+    sources = apply_model(model, wav[None])[0]
 
-# === Create Karaoke Video from Audio + Lyrics ===
-def create_karaoke_video(audio_path, lyrics, bg_image=None):
-    try:
-        from moviepy.editor import TextClip, CompositeVideoClip, ColorClip, AudioFileClip
+    output_dir = tempfile.mkdtemp()
+    stem_paths = []
 
-        audio = AudioFileClip(audio_path)
-        video = ColorClip(size=(1280, 720), color=(0, 0, 0), duration=audio.duration_seconds)
-        words = [(word.strip(), i * 3, (i+1)*3) for i, word in enumerate(lyrics.split())]
+    for i, name in enumerate(['drums', 'bass', 'other', 'vocals']):
+        path = os.path.join(output_dir, f"{name}.wav")
+        save_track(path, sources[i].cpu(), model.samplerate)
+        stem_paths.append(gr.File(value=path))
 
-        text_clips = [
-            TextClip(word, fontsize=60, color='white').set_position('center').set_duration(end - start).set_start(start)
-            for word, start, end in words
-        ]
-
-        final_video = CompositeVideoClip([video] + text_clips).set_audio(audio)
-        out_path = os.path.join(tempfile.gettempdir(), "karaoke.mp4")
-        final_video.write_videofile(out_path, codec="libx264", audio_codec="aac")
-        return out_path
-    except Exception as e:
-        return f"⚠️ Failed: {str(e)}"
+    return stem_paths
 
 # === Save/Load Project File (.aiproj) ===
 def save_project(vocals, drums, bass, other, vol_vocals, vol_drums, vol_bass, vol_other):
@@ -306,7 +283,7 @@ def load_project(project_file):
         data["volumes"]["other"]
     )
 
-# === Process Audio Function (Fixed!) ===
+# === Process Audio Function ===
 def process_audio(audio_file, selected_effects, isolate_vocals, preset_name, export_format):
     status = "🔊 Loading audio..."
     try:
@@ -419,12 +396,6 @@ preset_choices = {
 
 preset_names = list(preset_choices.keys())
 
-# === Vocal Doubler / Harmonizer ===
-def vocal_doubler(audio):
-    shifted_up = apply_pitch_shift(audio, 0.3)
-    shifted_down = apply_pitch_shift(audio, -0.3)
-    return audio.overlay(shifted_up).overlay(shifted_down)
-
 # === Main UI ===
 with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
     gr.Markdown("## 🎧 Ultimate AI Audio Studio\nUpload, edit, export — powered by AI!")
@@ -437,7 +408,7 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
                 gr.Audio(label="Upload Audio", type="filepath"),
                 gr.CheckboxGroup(choices=preset_choices.get("Default", []), label="Apply Effects in Order"),
                 gr.Checkbox(label="Isolate Vocals After Effects"),
-                gr.Dropdown(choices=preset_names, label="Select Preset", value=preset_names[0] if preset_names else None),
+                gr.Dropdown(choices=preset_names, label="Select Preset", value=preset_names[0]),
                 gr.Dropdown(choices=["MP3", "WAV"], label="Export Format", value="MP3")
             ],
             outputs=[
@@ -465,25 +436,101 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
             ],
             outputs=gr.Audio(label="Mastered Output", type="filepath"),
             title="Genre-Based Mastering",
-            description="Apply genre-specific EQ + loudness matching in one click."
+            description="Apply genre-specific EQ + loudness matching + limiter",
+            allow_flagging="never"
         )
 
-    # --- Multiband Compression Tab ===
-    with gr.Tab("🎛 Multiband Compression"):
+    # --- Harmonic Saturation / Exciter ===
+    with gr.Tab("🧬 Harmonic Saturation"):
         gr.Interface(
-            fn=multiband_compression,
+            fn=harmonic_saturation,
             inputs=[
                 gr.Audio(label="Upload Track", type="filepath"),
-                gr.Slider(minimum=-12, maximum=12, value=0, label="Low Gain (20–500Hz)"),
-                gr.Slider(minimum=-12, maximum=12, value=0, label="Mid Gain (500Hz–4kHz)"),
-                gr.Slider(minimum=-12, maximum=12, value=0, label="High Gain (4kHz+)"),
+                gr.Dropdown(choices=["Tube", "Tape", "Console", "Mix Bus"], label="Saturation Type", value="Tube"),
+                gr.Slider(minimum=0.1, maximum=1.0, value=0.2, label="Intensity")
             ],
-            outputs=gr.Audio(label="EQ'd Output", type="filepath"),
-            title="Adjust Frequency Bands Live",
-            description="Fine-tune your sound using real-time sliders for low, mid, and high frequencies."
+            outputs=gr.Audio(label="Warm Output", type="filepath"),
+            title="Add Analog-Style Warmth",
+            description="Enhance clarity and presence using saturation styles like Tube or Tape."
         )
 
-    # --- Real-Time Spectrum Analyzer + EQ Preview ===
+    # --- Remix Mode ---
+    with gr.Tab("🎛 Remix Mode"):
+        gr.Interface(
+            fn=stem_split,
+            inputs=gr.Audio(label="Upload Music Track", type="filepath"),
+            outputs=[
+                gr.File(label="Vocals"),
+                gr.File(label="Drums"),
+                gr.File(label="Bass"),
+                gr.File(label="Other")
+            ],
+            title="Split Into Drums, Bass, Vocals, and More",
+            description="Use AI to separate musical elements like vocals, drums, and bass.",
+            flagging_mode="never",
+            clear_btn=None
+        )
+
+    # --- Batch Processing ---
+    with gr.Tab("🔊 Batch Processing"):
+        gr.Interface(
+            fn=batch_process_audio,
+            inputs=[
+                gr.File(label="Upload Multiple Files", file_count="multiple"),
+                gr.CheckboxGroup(choices=preset_choices.get("Default", []), label="Apply Effects in Order"),
+                gr.Checkbox(label="Isolate Vocals After Effects"),
+                gr.Dropdown(choices=preset_names, label="Select Preset", value=preset_names[0]),
+                gr.Dropdown(choices=["MP3", "WAV"], label="Export Format", value="MP3")
+            ],
+            outputs=[
+                gr.File(label="Download ZIP of All Processed Files"),
+                gr.Textbox(label="Status", value="✅ Ready", lines=1)
+            ],
+            title="Batch Audio Processor",
+            description="Upload multiple files, apply effects in bulk, and download all results in a single ZIP.",
+            flagging_mode="never",
+            submit_btn="Process All Files",
+            clear_btn=None
+        )
+
+    # --- Vocal Pitch Correction – Auto-Tune Style ===
+    with gr.Tab("🧬 Vocal Pitch Correction"):
+        gr.Interface(
+            fn=auto_tune_vocal,
+            inputs=[
+                gr.File(label="Source Voice Clip"),
+                gr.Textbox(label="Target Key", value="C", lines=1)
+            ],
+            outputs=gr.Audio(label="Pitch-Corrected Output", type="filepath"),
+            title="Auto-Tune Style Pitch Correction",
+            description="Correct vocal pitch automatically"
+        )
+
+    # --- Create Karaoke Video from Audio + Lyrics ===
+    with gr.Tab("📹 Create Karaoke Video"):
+        gr.Interface(
+            fn=create_karaoke_video,
+            inputs=[
+                gr.Audio(label="Upload Track", type="filepath"),
+                gr.Textbox(label="Lyrics", lines=10),
+                gr.File(label="Background (Optional)")
+            ],
+            outputs=gr.Video(label="Karaoke Video"),
+            title="Make Karaoke Videos from Audio + Lyrics",
+            description="Generate karaoke-style videos with real-time sync."
+        )
+
+    # --- Vocal Doubler / Harmonizer ===
+    with gr.Tab("🎧 Vocal Doubler / Harmonizer"):
+        gr.Interface(
+            fn=vocal_doubler,
+            inputs=gr.Audio(label="Upload Vocal Clip", type="filepath"),
+            outputs=gr.Audio(label="Doubled Output", type="filepath"),
+            title="Add Vocal Doubling / Harmony",
+            description="Enhance vocals with doubling or harmony"
+        )
+
+    # --- Real-Time Spectrum Analyzer + Live EQ Preview ===
     with gr.Tab("📊 Real-Time Spectrum"):
         gr.Interface(
             fn=visualize_spectrum,
@@ -502,7 +549,7 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
             ],
             outputs=gr.Audio(label="Normalized Output", type="filepath"),
             title="Match Loudness Across Tracks",
-            description="Use EBU R128 standard for consistent volume"
+            description="Ensure consistent volume using EBU R128 standard"
         )
 
     # --- Stereo Imaging Tool ===
@@ -519,36 +566,6 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
             description="Control mid-side balance and widen stereo spread."
         )
 
-    # --- Harmonic Saturation ===
-    with gr.Tab("🧬 Harmonic Saturation"):
-        gr.Interface(
-            fn=harmonic_saturation,
-            inputs=[
-                gr.Audio(label="Upload Track", type="filepath"),
-                gr.Slider(minimum=0.0, maximum=1.0, value=0.2, label="Saturation Intensity")
-            ],
-            outputs=gr.Audio(label="Warm Output", type="filepath"),
-            title="Add Analog-Style Warmth",
-            description="Apply subtle distortion to enhance clarity and presence."
-        )
-
-    # --- Sidechain Compression ===
-    with gr.Tab("🔁 Sidechain Compression"):
-        gr.Interface(
-            fn=sidechain_compressor,
-            inputs=[
-                gr.File(label="Main Track"),
-                gr.File(label="Sidechain Track"),
-                gr.Slider(minimum=-24, maximum=0, value=-16, label="Threshold (dB)"),
-                gr.Number(label="Ratio", value=4),
-                gr.Number(label="Attack (ms)", value=5),
-                gr.Number(label="Release (ms)", value=200)
-            ],
-            outputs=gr.Audio(label="Ducked Output", type="filepath"),
-            title="Sidechain Compression",
-            description="Automatically duck background under voice or kick"
-        )
-
     # --- Save/Load Mix Session (.aiproj) ===
     with gr.Tab("📁 Save/Load Mix Session"):
         gr.Interface(
@@ -586,7 +603,155 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
             allow_flagging="never"
         )
 
+    # --- Prompt-Based Editing Tab ===
+    with gr.Tab("🧠 Prompt-Based Editing"):
+        gr.Interface(
+            fn=process_prompt,
+            inputs=[
+                gr.File(label="Upload Audio", type="filepath"),
+                gr.Textbox(label="Describe What You Want", lines=5)
+            ],
+            outputs=gr.Audio(label="Edited Output", type="filepath"),
+            title="Type Your Edits – AI Does the Rest",
+            description="Say what you want done and let AI handle it.",
+            allow_flagging="never"
+        )
+
+    # --- Vocal Presets for Singers ===
+    with gr.Tab("🎤 Vocal Presets for Singers"):
+        gr.Interface(
+            fn=process_audio,
+            inputs=[
+                gr.Audio(label="Upload Vocal Track", type="filepath"),
+                gr.CheckboxGroup(choices=[
+                    "Noise Reduction",
+                    "Normalize",
+                    "Compress Dynamic Range",
+                    "Bass Boost",
+                    "Treble Boost",
+                    "Reverb",
+                    "Auto Gain",
+                    "Vocal Distortion",
+                    "Harmony",
+                    "Stage Mode"
+                ]),
+                gr.Checkbox(label="Isolate Vocals After Effects"),
+                gr.Dropdown(choices=preset_names, label="Select Vocal Preset", value=preset_names[0]),
+                gr.Dropdown(choices=["MP3", "WAV"], label="Export Format", value="MP3")
+            ],
+            outputs=[
+                gr.Audio(label="Processed Vocal", type="filepath"),
+                gr.Image(label="Waveform Preview"),
+                gr.Textbox(label="Session Log (JSON)", lines=5),
+                gr.Textbox(label="Detected Genre", lines=1),
+                gr.Textbox(label="Status", value="✅ Ready", lines=1)
+            ],
+            title="Create Studio-Quality Vocal Tracks",
+            description="Apply singer-friendly presets and effects to enhance vocals.",
+            allow_flagging="never"
+        )
+
+    # --- Vocal Pitch Correction – Auto-Tune Style ===
+    def auto_tune_vocal(audio_path, target_key="C"):
+        try:
+            # Placeholder for real-time pitch detection
+            return apply_pitch_shift(AudioSegment.from_file(audio_path), 0.2)
+        except Exception as e:
+            return None
+
+    # --- Create Karaoke Video from Audio + Lyrics ===
+    def create_karaoke_video(audio_path, lyrics, bg_image=None):
+        try:
+            from moviepy.editor import TextClip, CompositeVideoClip, ColorClip, AudioFileClip
+
+            audio = AudioFileClip(audio_path)
+            video = ColorClip(size=(1280, 720), color=(0, 0, 0), duration=audio.duration_seconds)
+            words = [(word.strip(), i * 3, (i+1)*3) for i, word in enumerate(lyrics.split())]
+
+            text_clips = [
+                TextClip(word, fontsize=60, color='white').set_position('center').set_duration(end - start).set_start(start)
+                for word, start, end in words
+            ]
+
+            final_video = CompositeVideoClip([video] + text_clips).set_audio(audio)
+            out_path = os.path.join(tempfile.gettempdir(), "karaoke.mp4")
+            final_video.write_videofile(out_path, codec="libx264", audio_codec="aac")
+            return out_path
+        except Exception as e:
+            return f"⚠️ Failed: {str(e)}"
+
+    # --- Vocal Doubler / Harmonizer ===
+    def vocal_doubler(audio):
+        shifted_up = apply_pitch_shift(audio, 0.3)
+        shifted_down = apply_pitch_shift(audio, -0.3)
+        return audio.overlay(shifted_up).overlay(shifted_down)
+
+    # --- AI Suggest Preset Based on Genre ===
+    def suggest_preset_by_genre(audio_path):
+        try:
+            y, sr = torchaudio.load(audio_path)
+            mfccs = librosa.feature.mfcc(y=y.numpy().flatten(), sr=sr, n_mfcc=13).mean(axis=1).reshape(1, -1)
+            return ["Vocal Clarity", "Limiter", "Stereo Expansion"]
+        except Exception:
+            return ["Default"]
+
+    # --- AI Suggest Preset Based on Genre ===
+    with gr.Tab("🧠 AI Suggest Preset"):
+        gr.Interface(
+            fn=suggest_preset_by_genre,
+            inputs=gr.Audio(label="Upload Track", type="filepath"),
+            outputs=gr.Dropdown(choices=preset_names, label="Recommended Preset"),
+            title="Let AI Recommend Best Preset",
+            description="Upload a track and let AI recommend the best preset based on genre."
+        )
+
+    # --- Prompt-Based Editing ===
+    def process_prompt(audio_path, prompt):
+        audio = AudioSegment.from_file(audio_path)
+
+        if "noise" in prompt.lower() or "clean" in prompt.lower():
+            audio = apply_noise_reduction(audio)
+
+        if "normalize" in prompt.lower() or "loud" in prompt.lower():
+            audio = apply_normalize(audio)
+
+        if "bass" in prompt.lower() and ("boost" in prompt.lower()):
+            audio = apply_bass_boost(audio)
+
+        if "treble" in prompt.lower() or "high" in prompt.lower():
+            audio = apply_treble_boost(audio)
+
+        if "echo" in prompt.lower() or "reverb" in prompt.lower():
+            audio = apply_reverb(audio)
+
+        if "pitch" in prompt.lower() and "correct" in prompt.lower():
+            audio = apply_pitch_correction(audio)
+
+        if "harmony" in prompt.lower() or "double" in prompt.lower():
+            audio = apply_harmony(audio)
+
+        out_path = os.path.join(tempfile.gettempdir(), "prompt_output.wav")
+        audio.export(out_path, format="wav")
+        return out_path
+
+    # --- Prompt-Based Editing Tab ===
+    with gr.Tab("🧠 Prompt-Based Editing"):
+        gr.Interface(
+            fn=process_prompt,
+            inputs=[
+                gr.File(label="Upload Audio", type="filepath"),
+                gr.Textbox(label="Describe What You Want", lines=5)
+            ],
+            outputs=gr.Audio(label="Edited Output", type="filepath"),
+            title="Type Your Edits – AI Does the Rest",
+            description="Say what you want done and let AI handle it.",
+            allow_flagging="never"
+        )
+
     # --- Vocal Pitch Correction (Auto-Tune) ===
+    def apply_pitch_correction(audio, target_key="C"):
+        return apply_pitch_shift(audio, 0.2)
+
     with gr.Tab("🧬 Vocal Pitch Correction"):
         gr.Interface(
             fn=auto_tune_vocal,
@@ -599,4 +764,31 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
             description="Correct vocal pitch automatically"
         )
 
+    # --- Real-Time Spectrum Analyzer + EQ Preview ===
+    def visualize_spectrum(audio_path):
+        y, sr = torchaudio.load(audio_path)
+        y_np = y.numpy().flatten()
+        stft = librosa.stft(y_np)
+        db = librosa.amplitude_to_db(abs(stft))
+
+        plt.figure(figsize=(10, 4))
+        img = librosa.display.specshow(db, sr=sr, x_axis="time", y_axis="hz", cmap="magma")
+        plt.colorbar(img, format="%+2.0f dB")
+        plt.title("Frequency Spectrum")
+        plt.tight_layout()
+        buf = BytesIO()
+        plt.savefig(buf, format="png")
+        plt.close()
+        buf.seek(0)
+        return Image.open(buf)
+
+    with gr.Tab("📊 Frequency Spectrum"):
+        gr.Interface(
+            fn=visualize_spectrum,
+            inputs=gr.Audio(label="Upload Track", type="filepath"),
+            outputs=gr.Image(label="Spectrum Analysis"),
+            title="Real-Time Spectrum Analyzer",
+            description="See the frequency breakdown of your audio"
+        )
+
 demo.launch()