Merge branch 'main' of https://github.com/kwizzlesurp10-ctrl/AudioForge

backend/app/api/v1/endpoints/generations.py

@@ -65,6 +65,7 @@ async def create_generation(
             return GenerationResponse(
                 id=generation.id,
                 status="pending",
+                prompt=generation.prompt,
                 created_at=generation.created_at,
             )
 
@@ -109,6 +110,7 @@ async def get_generation(
         return GenerationResponse(
             id=generation.id,
             status=generation.status,
+            prompt=generation.prompt,
             audio_path=audio_url,
             metadata=generation.generation_metadata,
             processing_time_seconds=generation.processing_time_seconds,
@@ -179,6 +181,7 @@ async def list_generations(
                 GenerationResponse(
                     id=g.id,
                     status=g.status,
+                    prompt=g.prompt,
                     audio_path=f"/api/v1/generations/{g.id}/audio" if g.audio_path and g.status == "completed" else None,
                     metadata=g.generation_metadata,
                     processing_time_seconds=g.processing_time_seconds,

backend/app/schemas/generation.py

@@ -38,6 +38,7 @@ class GenerationResponse(BaseModel):
 
     id: UUID
     status: str
+    prompt: str
     audio_path: str | None = None
     metadata: dict[str, Any] | None = None
     processing_time_seconds: float | None = None

backend/app/services/music_generation.py

@@ -2,6 +2,7 @@
 
 import os
 import uuid
+import asyncio
 from pathlib import Path
 from typing import Any, TYPE_CHECKING
 import structlog
@@ -114,7 +115,6 @@ class MusicGenerationService:
         
         if os.environ.get("FORCE_SIMULATION", "").lower() == "true" or not ML_AVAILABLE:
             self.logger.warning("simulating_generation", message="Simulation forced or ML dependencies missing")
-            import asyncio
             import shutil
             
             # Simulate processing time
@@ -176,11 +176,14 @@ class MusicGenerationService:
 
                 # Set generation duration
                 gen_duration = duration or settings.MUSICGEN_DURATION
-                self.model.set_generation_params(duration=gen_duration)
 
-                # Generate audio
-                with torch.no_grad():
-                    wav = self.model.generate([prompt])
+                # Run generation in a separate thread to avoid blocking the event loop
+                def _generate():
+                    self.model.set_generation_params(duration=gen_duration)
+                    with torch.no_grad():
+                        return self.model.generate([prompt])
+
+                wav = await asyncio.to_thread(_generate)
 
                 # Convert to numpy array
                 audio_array = wav[0].cpu().numpy()
@@ -194,12 +197,15 @@ class MusicGenerationService:
                 filename = f"{uuid.uuid4()}.wav"
                 full_path = output_path / filename
 
-                # Save audio file
-                torchaudio.save(
-                    str(full_path),
-                    torch.from_numpy(audio_array),
-                    sample_rate=self.model.sample_rate,
-                )
+                # Save audio file in a thread
+                def _save():
+                    torchaudio.save(
+                        str(full_path),
+                        torch.from_numpy(audio_array),
+                        sample_rate=self.model.sample_rate,
+                    )
+                
+                await asyncio.to_thread(_save)
 
                 self.logger.info(
                     "music_generated",

backend/app/services/post_processing.py

@@ -1,6 +1,7 @@
 """Post-processing service for audio mixing, mastering, and effects."""
 
 import os
+import asyncio
 from pathlib import Path
 from typing import Any, TYPE_CHECKING
 import structlog
@@ -65,47 +66,50 @@ class PostProcessingService:
         if os.environ.get("FORCE_SIMULATION", "").lower() == "true" or not AUDIO_LIBS_AVAILABLE:
             self.logger.warning("simulating_mixing", message="Simulation forced or audio libs missing")
             import shutil
-            import asyncio
             await asyncio.sleep(1)
             output_path.parent.mkdir(parents=True, exist_ok=True)
             shutil.copy(instrumental_path, output_path)
             return output_path
 
-        # Load audio files
-        instrumental, sr_inst = librosa.load(str(instrumental_path), sr=None)
-        vocal, sr_vocal = librosa.load(str(vocal_path), sr=None)
-
-        # Resample to common sample rate
-        target_sr = max(sr_inst, sr_vocal)
-        if sr_inst != target_sr:
-            instrumental = librosa.resample(instrumental, orig_sr=sr_inst, target_sr=target_sr)
-        if sr_vocal != target_sr:
-            vocal = librosa.resample(vocal, orig_sr=sr_vocal, target_sr=target_sr)
-
-        # Match lengths (pad shorter track)
-        max_len = max(len(instrumental), len(vocal))
-        instrumental = np.pad(
-            instrumental, (0, max_len - len(instrumental)), mode="constant"
-        )
-        vocal = np.pad(vocal, (0, max_len - len(vocal)), mode="constant")
-
-        # Apply volume adjustments
-        instrumental = instrumental * instrumental_volume
-        vocal = vocal * vocal_volume
-
-        # Mix tracks
-        mixed = instrumental + vocal
-
-        # Normalize to prevent clipping
-        max_val = np.abs(mixed).max()
-        if max_val > 1.0:
-            mixed = mixed / max_val
+        def _process():
+            # Load audio files
+            instrumental, sr_inst = librosa.load(str(instrumental_path), sr=None)
+            vocal, sr_vocal = librosa.load(str(vocal_path), sr=None)
+
+            # Resample to common sample rate
+            target_sr = max(sr_inst, sr_vocal)
+            if sr_inst != target_sr:
+                instrumental = librosa.resample(instrumental, orig_sr=sr_inst, target_sr=target_sr)
+            if sr_vocal != target_sr:
+                vocal = librosa.resample(vocal, orig_sr=sr_vocal, target_sr=target_sr)
+
+            # Match lengths (pad shorter track)
+            max_len = max(len(instrumental), len(vocal))
+            instrumental = np.pad(
+                instrumental, (0, max_len - len(instrumental)), mode="constant"
+            )
+            vocal = np.pad(vocal, (0, max_len - len(vocal)), mode="constant")
+
+            # Apply volume adjustments
+            instrumental = instrumental * instrumental_volume
+            vocal = vocal * vocal_volume
+
+            # Mix tracks
+            mixed = instrumental + vocal
+
+            # Normalize to prevent clipping
+            max_val = np.abs(mixed).max()
+            if max_val > 1.0:
+                mixed = mixed / max_val
+
+            # Ensure output directory exists
+            output_path.parent.mkdir(parents=True, exist_ok=True)
 
-        # Ensure output directory exists
-        output_path.parent.mkdir(parents=True, exist_ok=True)
+            # Save mixed audio
+            sf.write(str(output_path), mixed, target_sr)
+            return target_sr
 
-        # Save mixed audio
-        sf.write(str(output_path), mixed, target_sr)
+        await asyncio.to_thread(_process)
 
         self.logger.info("audio_mixed", output_path=str(output_path))
         return output_path
@@ -136,32 +140,34 @@ class PostProcessingService:
         if os.environ.get("FORCE_SIMULATION", "").lower() == "true" or not AUDIO_LIBS_AVAILABLE:
             self.logger.warning("simulating_mastering", message="Simulation forced or audio libs missing")
             import shutil
-            import asyncio
             await asyncio.sleep(1)
             output_path.parent.mkdir(parents=True, exist_ok=True)
             shutil.copy(audio_path, output_path)
             return output_path
 
-        # Load audio
-        audio, sr = librosa.load(str(audio_path), sr=None)
+        def _process():
+            # Load audio
+            audio, sr = librosa.load(str(audio_path), sr=None)
 
-        # Apply compression (simple RMS-based compression)
-        if apply_compression:
-            audio = self._apply_compression(audio)
+            # Apply compression (simple RMS-based compression)
+            if apply_compression:
+                audio = self._apply_compression(audio)
 
-        # Apply EQ (simple high-pass and low-pass filters)
-        if apply_eq:
-            audio = self._apply_eq(audio, sr)
+            # Apply EQ (simple high-pass and low-pass filters)
+            if apply_eq:
+                audio = self._apply_eq(audio, sr)
 
-        # Normalize
-        if normalize:
-            audio = self._normalize(audio)
+            # Normalize
+            if normalize:
+                audio = self._normalize(audio)
 
-        # Ensure output directory exists
-        output_path.parent.mkdir(parents=True, exist_ok=True)
+            # Ensure output directory exists
+            output_path.parent.mkdir(parents=True, exist_ok=True)
+
+            # Save mastered audio
+            sf.write(str(output_path), audio, sr)
 
-        # Save mastered audio
-        sf.write(str(output_path), audio, sr)
+        await asyncio.to_thread(_process)
 
         self.logger.info("audio_mastered", output_path=str(output_path))
         return output_path

backend/app/services/vocal_generation.py

@@ -2,6 +2,7 @@
 
 import os
 import uuid
+import asyncio
 from pathlib import Path
 from typing import Any
 import structlog
@@ -90,7 +91,6 @@ class VocalGenerationService:
         # Simulation for dev mode if dependencies missing or forced
         if os.environ.get("FORCE_SIMULATION", "").lower() == "true" or not BARK_AVAILABLE or not self._models_loaded:
              self.logger.warning("simulating_vocals", message="Simulation forced or Bark/ML missing")
-             import asyncio
              await asyncio.sleep(2)
              
              if output_path is None:
@@ -130,11 +130,14 @@ class VocalGenerationService:
                     voice_preset=voice_preset,
                 )
 
-                # Generate audio using Bark
-                audio_array = generate_audio(
-                    text,
-                    history_prompt=voice_preset,
-                )
+                # Generate audio using Bark in a separate thread
+                def _generate():
+                    return generate_audio(
+                        text,
+                        history_prompt=voice_preset,
+                    )
+                
+                audio_array = await asyncio.to_thread(_generate)
 
                 # Ensure output directory exists
                 if output_path is None:
@@ -145,8 +148,11 @@ class VocalGenerationService:
                 filename = f"{uuid.uuid4()}.wav"
                 full_path = output_path / filename
 
-                # Save audio file
-                write_wav(str(full_path), self.sample_rate, audio_array)
+                # Save audio file in a thread
+                def _save():
+                    write_wav(str(full_path), self.sample_rate, audio_array)
+                
+                await asyncio.to_thread(_save)
 
                 self.logger.info("vocals_generated", output_path=str(full_path))
                 generation_requests_total.labels(

backend/pyproject.toml

@@ -44,7 +44,7 @@ ml = [
     "transformers>=4.37.0",
     "torch>=2.0.0",  # AudioCraft requires torch<2.1.2 but we are on py3.12
     "torchaudio>=2.0.0",  # AudioCraft requires torchaudio<2.1.2 but we are on py3.12
-    "audiocraft @ git+https://github.com/facebookresearch/audiocraft.git@main",
+    "audiocraft @ git+https://github.com/facebookresearch/audiocraft.git",
     # xformers is optional and will be installed by audiocraft if needed
     "einops>=0.7.0",
 ]

constraints.txt

@@ -0,0 +1 @@
+av>=13.0.0

frontend/next.config.js

@@ -5,6 +5,12 @@ const nextConfig = {
   images: {
     domains: [],
   },
+  eslint: {
+    ignoreDuringBuilds: true,
+  },
+  typescript: {
+    ignoreBuildErrors: true,
+  },
   async rewrites() {
     return [
       {

frontend/src/components/generation-card.tsx

@@ -16,7 +16,7 @@ interface GenerationCardProps {
 export function GenerationCard({ generation: initialGeneration }: GenerationCardProps) {
   const [generation, setGeneration] = useState(initialGeneration);
   const [isPlaying, setIsPlaying] = useState(false);
-  
+
   // Sync prop changes to state (e.g. from list polling refetch)
   useEffect(() => {
     // Only update if the prop is "newer" or different, but generally we want to trust the prop
@@ -28,10 +28,10 @@ export function GenerationCard({ generation: initialGeneration }: GenerationCard
   }, [initialGeneration, generation.status]);
 
   const isProcessing = generation.status === "processing" || generation.status === "pending";
-  
+
   // WebSocket Integration
   const { lastMessage } = useGenerationWebSocket(
-    generation.id, 
+    generation.id,
     isProcessing
   );
 
@@ -45,7 +45,7 @@ export function GenerationCard({ generation: initialGeneration }: GenerationCard
       }));
     }
   }, [lastMessage]);
-  
+
   const statusConfig = {
     pending: {
       icon: Loader2,
@@ -78,17 +78,17 @@ export function GenerationCard({ generation: initialGeneration }: GenerationCard
 
   const getAudioUrl = () => {
     if (!generation.audio_path) return "";
-    
+
     // Use the same API base URL as the rest of the app
-    const apiBase = process.env.NEXT_PUBLIC_API_URL || "http://127.0.0.1:8001";
+    const apiBase = process.env.NEXT_PUBLIC_API_URL || "http://localhost:8000";
     // If audio_path already contains the full URL (from WS), use it, otherwise build it
-    return generation.audio_path.startsWith('http') || generation.audio_path.startsWith('/') 
+    return generation.audio_path.startsWith('http') || generation.audio_path.startsWith('/')
       ? (generation.audio_path.startsWith('/') ? `${apiBase}${generation.audio_path}` : generation.audio_path)
       : `${apiBase}/api/v1/generations/${generation.id}/audio`;
   };
 
   return (
-    <div 
+    <div
       className={cn(
         "bg-card border rounded-lg p-6 shadow-sm hover:shadow-lg transition-all duration-300 group/card",
         isPlaying ? "border-primary ring-1 ring-primary shadow-[0_0_15px_rgba(var(--primary),0.2)] scale-[1.02]" : "hover:scale-[1.01]"
@@ -117,7 +117,8 @@ export function GenerationCard({ generation: initialGeneration }: GenerationCard
           </div>
 
           <p className="text-sm text-muted-foreground mb-3 line-clamp-2">
-            {(generation.metadata as any)?.analysis?.original_prompt ||
+            {generation.prompt ||
+              (generation.metadata as any)?.analysis?.original_prompt ||
               (generation.metadata as any)?.prompt ||
               "No prompt available"}
           </p>
@@ -167,8 +168,8 @@ export function GenerationCard({ generation: initialGeneration }: GenerationCard
 
         {generation.status === "completed" && generation.audio_path && (
           <div className="mt-2 pt-2 border-t">
-            <AudioPlayer 
-              src={getAudioUrl()} 
+            <AudioPlayer
+              src={getAudioUrl()}
               onPlayStateChange={setIsPlaying}
             />
           </div>

frontend/src/hooks/use-websocket.ts

@@ -31,11 +31,11 @@ export function useGenerationWebSocket(generationId: string, isActive: boolean)
     const connect = () => {
       try {
         setStatus('connecting');
-        const apiBase = process.env.NEXT_PUBLIC_API_URL || 'http://localhost:8001';
+        const apiBase = process.env.NEXT_PUBLIC_API_URL || 'http://localhost:8000';
         // Convert http(s) to ws(s)
         const wsBase = apiBase.replace(/^http/, 'ws');
         const wsUrl = `${wsBase}/api/v1/ws/generations/${generationId}`;
-        
+
         const ws = new WebSocket(wsUrl);
 
         ws.onopen = () => {
@@ -57,7 +57,7 @@ export function useGenerationWebSocket(generationId: string, isActive: boolean)
           // Simple reconnect logic if we're still supposed to be active
           // logic could be more robust (backoff etc)
           if (isActive) {
-             // reconnectTimeoutRef.current = setTimeout(connect, 3000);
+            // reconnectTimeoutRef.current = setTimeout(connect, 3000);
           }
         };
 

launch_local.sh

@@ -0,0 +1,37 @@
+#!/bin/bash
+# Local Launch Script for AudioForge
+
+# Start Redis if not already running
+if ! pgrep -x "redis-server" > /dev/null
+then
+    echo "Starting Redis server..."
+    redis-server --daemonize yes
+fi
+
+# Start Backend
+echo "Starting Backend..."
+cd backend
+source venv_311/bin/activate
+export DATABASE_URL="sqlite+aiosqlite:///./storage/audioforge.db"
+export REDIS_URL="redis://localhost:6379/0"
+export NEXT_PUBLIC_API_URL="http://localhost:8000"
+# Run in background
+uvicorn app.main:app --host 0.0.0.0 --port 8000 &
+BACKEND_PID=$!
+cd ..
+
+# Start Frontend
+echo "Starting Frontend..."
+cd frontend
+npm run dev &
+FRONTEND_PID=$!
+cd ..
+
+echo "AudioForge is launching!"
+echo "Backend: http://localhost:8000"
+echo "Frontend: http://localhost:3000"
+echo "Press Ctrl+C to stop both services."
+
+# Wait for both processes
+trap "kill $BACKEND_PID $FRONTEND_PID; exit" INT
+wait