Create app.py

app.py

@@ -0,0 +1,178 @@
+import os
+import gradio as gr
+import torch
+import numpy as np
+import librosa
+import soundfile as sf
+import traceback
+import tempfile
+
+# -----------------------------
+# MODEL PATHS
+# -----------------------------
+MODEL_DIR = "models"
+PTH_PATH = os.path.join(MODEL_DIR, "elvish.pth")
+INDEX_PATH = os.path.join(MODEL_DIR, "elvish.index")
+
+# -----------------------------
+# SAFE DEVICE
+# -----------------------------
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+# -----------------------------
+# LOAD MODEL (SAFE WRAPPER)
+# -----------------------------
+model = None
+
+def load_model():
+    global model
+    try:
+        if not os.path.exists(PTH_PATH):
+            print("❌ Model .pth missing")
+            return None
+
+        # NOTE: Real RVC model loader depends on repo
+        # This is safe placeholder loader for HF Spaces stability
+        model = torch.load(PTH_PATH, map_location=device)
+        print("✅ Model loaded successfully")
+        return model
+
+    except Exception as e:
+        print("❌ Model loading failed:", e)
+        return None
+
+model = load_model()
+
+# -----------------------------
+# AUDIO PROCESSING HELPERS
+# -----------------------------
+def load_audio(file_path):
+    try:
+        audio, sr = librosa.load(file_path, sr=16000)
+        return audio, sr
+    except Exception:
+        return None, None
+
+# -----------------------------
+# SAFE RVC CONVERSION CORE
+# -----------------------------
+def voice_conversion(audio_path):
+    try:
+        if model is None:
+            return None, "❌ Model not loaded"
+
+        audio, sr = load_audio(audio_path)
+        if audio is None:
+            return None, "❌ Invalid audio file"
+
+        # -----------------------------
+        # FAKE SAFE TRANSFORM (PLACEHOLDER)
+        # -----------------------------
+        # Real RVC logic goes here (inference pipeline)
+        processed_audio = audio * 1.0  # no crash safe pass-through
+
+        # save output
+        out_path = tempfile.mktemp(suffix=".wav")
+        sf.write(out_path, processed_audio, sr)
+
+        return out_path, "✅ Conversion done"
+
+    except Exception as e:
+        return None, f"❌ Error: {str(e)}"
+
+# -----------------------------
+# VIDEO → AUDIO EXTRACTION
+# -----------------------------
+def video_to_audio(video_file):
+    try:
+        import moviepy.editor as mp
+
+        clip = mp.VideoFileClip(video_file)
+        out_path = tempfile.mktemp(suffix=".wav")
+        clip.audio.write_audiofile(out_path)
+
+        return out_path, "✅ Video converted to audio"
+
+    except Exception as e:
+        return None, f"❌ Video error: {str(e)}"
+
+# -----------------------------
+# WRAPPERS FOR UI
+# -----------------------------
+def voice_to_voice(file):
+    try:
+        out, msg = voice_conversion(file)
+        return out, msg
+    except:
+        return None, "❌ Unexpected error"
+
+def video_to_voice(file):
+    try:
+        audio_path, msg = video_to_audio(file)
+        if audio_path is None:
+            return None, msg
+
+        out, msg2 = voice_conversion(audio_path)
+        return out, msg2
+
+    except Exception:
+        return None, "❌ Video-to-voice failed"
+
+# -----------------------------
+# TEXT TO VOICE (SAFE MOCK)
+# -----------------------------
+def text_to_voice(text):
+    try:
+        # NOTE: RVC doesn't generate speech itself
+        # So we simulate safe fallback or TTS hook
+        import pyttsx3
+
+        engine = pyttsx3.init()
+        out_path = tempfile.mktemp(suffix=".wav")
+        engine.save_to_file(text, out_path)
+        engine.runAndWait()
+
+        return out_path, "✅ Text converted (TTS fallback)"
+
+    except Exception as e:
+        return None, f"❌ TTS error: {str(e)}"
+
+# -----------------------------
+# GRADIO UI
+# -----------------------------
+with gr.Blocks() as app:
+
+    gr.Markdown("# 🎤 Elvish AI Voice System (RVC)")
+
+    with gr.Tab("Text → Voice"):
+        t_input = gr.Textbox(label="Enter Text")
+        t_btn = gr.Button("Convert")
+        t_out_audio = gr.Audio()
+        t_status = gr.Textbox()
+
+        t_btn.click(text_to_voice, t_input, [t_out_audio, t_status])
+
+    with gr.Tab("Voice → Voice"):
+        v_input = gr.Audio(type="filepath")
+        v_btn = gr.Button("Convert Voice")
+        v_out_audio = gr.Audio()
+        v_status = gr.Textbox()
+
+        v_btn.click(voice_to_voice, v_input, [v_out_audio, v_status])
+
+    with gr.Tab("Video → Voice"):
+        vid_input = gr.Video()
+        vid_btn = gr.Button("Convert Video")
+        vid_out_audio = gr.Audio()
+        vid_status = gr.Textbox()
+
+        vid_btn.click(video_to_voice, vid_input, [vid_out_audio, vid_status])
+
+# -----------------------------
+# RUN APP
+# -----------------------------
+app.launch(
+    server_name="0.0.0.0",
+    server_port=7860,
+    debug=True
+)