Update app.py

app.py

@@ -127,22 +127,97 @@ def apply_stage_mode(audio):
     processed = apply_bass_boost(processed, gain=6)
     return apply_limiter(processed, limit_dB=-2)
 
-# === Genre Mastering Presets ===
-genre_presets = {
-    "Soul": ["Noise Reduction", "Bass Boost (+6dB)", "Mid Enhance"],
-    "Funk": ["Treble Boost (+6dB)", "Compression", "Stereo Widening"],
-    "Rock": ["Distortion", "Punchy Mids", "Reverb"],
-    "Pop": ["Vocal Clarity", "Limiter", "Stereo Expansion"],
-    "Acoustic": ["Natural Reverb", "Gentle Compression", "Mid Focus"],
-    "Dance": ["Loudness Maximizer", "Bass Emphasis", "Stereo Widen"],
-    "EDM": ["Heavy Bass", "Stereo Expansion", "Limiter", "Phaser"],
-    "Country": ["Clean Mix", "Subtle Reverb", "Mid Focus"],
-    "Disco": ["Rhythmic Echo", "Bass Thump", "Treble Boost (+8dB)"],
-    "Metal": ["Distortion", "High Gain", "Crisp Highs"],
-    "Hip-Hop": ["Deep Bass", "Vocal Presence", "Saturation"],
-    "Trap": ["808 Bass", "Reverb", "Lo-Fi Texture"],
-    "LoFi": ["Bitcrusher", "Tape Hiss", "Soft Compression"]
-}
+# === Auto-EQ per Genre ===
+def auto_eq(audio, genre="Pop"):
+    # Define frequency bands based on genre
+    eq_map = {
+        "Pop": [(200, 500, -3), (2000, 4000, +4)],  # Cut muddiness, boost vocals
+        "EDM": [(60, 250, +6), (8000, 12000, +3)],  # Maximize bass & sparkle
+        "Rock": [(1000, 3000, +4), (7000, 10000, -3)],  # Punchy mids, reduce sibilance
+        "Hip-Hop": [(20, 100, +6), (7000, 10000, -4)],  # Deep lows, smooth highs
+        "Acoustic": [(100, 300, -3), (4000, 8000, +2)],  # Natural tone
+        "Metal": [(100, 500, -4), (2000, 5000, +6), (7000, 12000, -3)],  # Clear low-mids, crisp highs
+        "Trap": [(80, 120, +6), (3000, 6000, -4)],  # Sub-bass boost, cut harsh highs
+        "LoFi": [(20, 200, +3), (1000, 3000, -2)],  # Warmth, soft mids
+        "Default": []
+    }
+
+    from scipy.signal import butter, sosfilt
+
+    def band_eq(samples, sr, lowcut, highcut, gain):
+        sos = butter(10, [lowcut, highcut], btype='band', output='sos', fs=sr)
+        filtered = sosfilt(sos, samples)
+        return samples + gain * filtered
+
+    samples, sr = audiosegment_to_array(audio)
+    samples = samples.astype(np.float64)
+
+    for band in eq_map.get(genre, []):
+        low, high, gain = band
+        samples = band_eq(samples, sr, low, high, gain)
+
+    return array_to_audiosegment(samples.astype(np.int16), sr, channels=audio.channels)
+
+# === AI Voice Effects – Harmony / Doubling / Tuning ===
+def pitch_correct(audio, target_key="C", semitones=None):
+    if semitones is None:
+        # Detect key and calculate needed shift
+        semitones = 0  # Placeholder
+    return apply_pitch_shift(audio, semitones)
+
+def vocal_doubling(audio):
+    double1 = apply_pitch_shift(audio, 0.3)
+    double2 = apply_pitch_shift(audio, -0.3)
+    return audio.overlay(double1).overlay(double2)
+
+# === Prompt-Based Editing ===
+def process_prompt(audio_path, prompt):
+    prompt = prompt.lower()
+    audio = AudioSegment.from_file(audio_path)
+
+    if "noise" in prompt or "clean" in prompt:
+        audio = apply_noise_reduction(audio)
+
+    if "normalize" in prompt or "loud" in prompt:
+        audio = apply_normalize(audio)
+
+    if "bass" in prompt and ("boost" in prompt or "up" in prompt):
+        audio = apply_bass_boost(audio)
+
+    if "treble" in prompt or "highs" in prompt:
+        audio = apply_treble_boost(audio)
+
+    if "echo" in prompt or "reverb" in prompt:
+        audio = apply_reverb(audio)
+
+    if "pitch" in prompt and "correct" in prompt:
+        audio = pitch_correct(audio)
+
+    if "harmony" in prompt or "double" in prompt:
+        audio = vocal_doubling(audio)
+
+    out_path = os.path.join(tempfile.gettempdir(), "prompt_output.wav")
+    audio.export(out_path, format="wav")
+    return out_path
+
+# === Spectrum Analyzer + EQ Visualizer ===
+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)
 
 # === Vocal Isolation Helpers ===
 def load_track_local(path, sample_rate, channels=2):
@@ -358,7 +433,7 @@ def transcribe_audio(audio_path):
 # === TTS Tab ===
 tts = TTS(model_name="tts_models/en/ljspeech/tacotron2-DDC", progress_bar=False)
 
-def generate_tts(text):
+def generate_tTS(text):
     out_path = os.path.join(tempfile.gettempdir(), "tts_output.wav")
     tts.tts_to_file(text=text, file_path=out_path)
     return out_path
@@ -407,12 +482,6 @@ def mix_tracks(track1, track2, volume_offset=0):
     mixed.export(out_path, format="wav")
     return out_path
 
-# === Genre Mastering Tab ===
-def apply_genre_preset(audio, genre):
-    global preset_choices
-    selected_preset = genre_presets.get(genre, [])
-    return process_audio(audio, selected_preset, False, genre, "WAV")[0]
-
 # === Dummy Voice Cloning Tab – Works Locally Only ===
 def clone_voice(*args):
     return "⚠️ Voice cloning requires local install – use Python 3.9 or below"
@@ -440,7 +509,6 @@ def diarize_and_transcribe(audio_path):
     audio.export(temp_wav, format="wav")
 
     try:
-        from pyannote.audio import Pipeline as DiarizationPipeline
         diarization = diarize_pipeline(temp_wav)
 
         result = whisper.transcribe(temp_wav)
@@ -517,7 +585,7 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
                 gr.File(label="Upload Multiple Files", file_count="multiple"),
                 gr.CheckboxGroup(choices=effect_options, 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=[
@@ -558,17 +626,32 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
             ],
             outputs=gr.Audio(label="Mastered Output", type="filepath"),
             title="Genre-Specific Mastering",
-            description="Apply professionally tuned presets for popular music genres."
+            description="Apply professionally tuned mastering settings for popular music genres."
+        )
+
+    # --- 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"
         )
 
-    # --- Transcribe & Edit Tab ===
-    with gr.Tab("📝 Transcribe & Edit"):
+    # --- Spectrum Analyzer Tab ===
+    with gr.Tab("📊 Frequency Spectrum"):
         gr.Interface(
-            fn=transcribe_audio,
-            inputs=gr.Audio(label="Upload Audio", type="filepath"),
-            outputs=gr.Textbox(label="Transcribed Text", lines=10),
-            title="Transcribe & Edit Spoken Content",
-            description="Convert voice to text and edit it before exporting again."
+            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",
+            allow_flagging="never"
         )
 
     # --- Voice Cloning (Local Only) ===
@@ -655,29 +738,4 @@ with gr.Blocks(title="AI Audio Studio", css="style.css") as demo:
             description="Detect and trim silence at start/end or between words"
         )
 
-    # --- Load/Save Project File (.aiproj) ===
-    with gr.Tab("📁 Save/Load Project"):
-        gr.Interface(
-            fn=save_project,
-            inputs=[
-                gr.File(label="Original Audio"),
-                gr.Dropdown(choices=preset_names, label="Used Preset", value=preset_names[0]),
-                gr.CheckboxGroup(choices=effect_options, label="Applied Effects")
-            ],
-            outputs=gr.File(label="Project File (.aiproj)"),
-            title="Save Everything Together",
-            description="Save your session, effects, and settings in one file to reuse later."
-        )
-
-        gr.Interface(
-            fn=load_project,
-            inputs=gr.File(label="Upload .aiproj File"),
-            outputs=[
-                gr.Dropdown(choices=preset_names, label="Loaded Preset"),
-                gr.CheckboxGroup(choices=effect_options, label="Loaded Effects")
-            ],
-            title="Resume Last Project",
-            description="Load your saved session"
-        )
-
 demo.launch()