Update app.py

app.py

@@ -1,91 +1,117 @@
-from fastapi import FastAPI, HTTPException, Request
-from fastapi.responses import FileResponse
+# app.py
+import os, io, time, uuid, shutil, tempfile
 from pathlib import Path
-import uuid
+from fastapi import FastAPI, Request
+from fastapi.responses import FileResponse, StreamingResponse, JSONResponse
+from fastapi.websockets import WebSocket
+from pydantic import BaseModel
 import subprocess
-import os
-import shutil
+import wave
 
 app = FastAPI()
 
-# ========== Dynamic VOICES_DIR fix ==========
-# Try /tmp/tts_app/voices first
-base_dirs = [Path("/tmp/tts_app/voices"), Path("/data/voices"), Path("/home/user/voices"), Path.cwd() / "voices"]
-
-for path in base_dirs:
-    try:
-        path.mkdir(parents=True, exist_ok=True)
-        test_file = path / ".write_test"
-        test_file.write_text("ok")
-        test_file.unlink()
-        VOICES_DIR = path
-        break
-    except Exception:
-        continue
-else:
-    raise RuntimeError("❌ Could not find any writable directory for VOICES_DIR.")
-
-FILES_DIR = VOICES_DIR.parent / "files"
+# ========== CONFIG ==========
+ROOT_DIR = Path(_file_).parent.resolve()
+VOICES_DIR = ROOT_DIR / "voices"
+FILES_DIR = ROOT_DIR / "files"
+VOICES_DIR.mkdir(parents=True, exist_ok=True)
 FILES_DIR.mkdir(parents=True, exist_ok=True)
+DEFAULT_VOICE = "en_US-libritts-high"  # Replace with actual voice file
+DEFAULT_SR = 22050
 
-# ========== Piper engine config ==========
-VOICE_TAGS = [
-    "en_US-libritts-high",   # Humanlike, CPU-friendly (VITS-based)
-    "en_US-amy-medium",      # Classic Piper Amy
-    "en_US-lessac-high",     # LJSpeech-derived VITS
-]
-
+# ========== HEALTH ==========
 @app.get("/health")
-def health():
+async def health():
     return {
         "ok": True,
         "engine": "piper-tts (CLI, CPU)",
+        "default_voice": DEFAULT_VOICE,
         "voice_dir": str(VOICES_DIR),
+        "available_voices": [f.stem for f in VOICES_DIR.glob("*.onnx")],
         "files_dir": str(FILES_DIR),
-        "default_voice": None,
-        "available_voices": VOICE_TAGS,
     }
 
+# ========== SPEAK (HTTP) ==========
+class SpeakRequest(BaseModel):
+    text: str
+    voice: str = DEFAULT_VOICE
+    rate_wpm: int = 170
+    length_scale: float = 1.0
+    noise_scale: float = 0.33
+    noise_w: float = 0.5
+
 @app.post("/speak")
-async def speak(request: Request):
+async def speak(req: SpeakRequest):
+    out_path = FILES_DIR / f"{uuid.uuid4().hex}.wav"
+    voice_path = VOICES_DIR / f"{req.voice}.onnx"
+
+    if not voice_path.exists():
+        return JSONResponse({"error": "Voice not found."}, status_code=404)
+
+    cmd = [
+        "piper",
+        "--model", str(voice_path),
+        "--output_file", str(out_path),
+        "--text", req.text,
+        "--length_scale", str(req.length_scale),
+        "--noise_scale", str(req.noise_scale),
+        "--noise_w", str(req.noise_w),
+    ]
+    subprocess.run(cmd, check=True)
+
+    return FileResponse(out_path, media_type="audio/wav")
+
+# ========== STREAM (WebSocket) ==========
+@app.websocket("/ws/tts")
+async def tts_stream(websocket: WebSocket):
+    await websocket.accept()
+    voice = DEFAULT_VOICE
+    settings = {
+        "length_scale": 1.0,
+        "noise_scale": 0.33,
+        "noise_w": 0.5
+    }
+
+    temp_file = FILES_DIR / f"{uuid.uuid4().hex}.wav"
+    wave_writer = wave.open(str(temp_file), 'wb')
+    wave_writer.setnchannels(1)
+    wave_writer.setsampwidth(2)
+    wave_writer.setframerate(DEFAULT_SR)
+
     try:
-        payload = await request.json()
-        text = payload.get("text", "").strip()
-        voice = payload.get("voice", VOICE_TAGS[0])
-        length_scale = float(payload.get("length_scale", 1.08))
-        noise_scale = float(payload.get("noise_scale", 0.33))
-        noise_w = float(payload.get("noise_w", 0.8))
-
-        if not text:
-            raise HTTPException(status_code=400, detail="No text provided")
-
-        output_file = FILES_DIR / f"{uuid.uuid4().hex}.wav"
-        command = [
-            "piper",
-            "--model", f"voices/{voice}.onnx",
-            "--output_file", str(output_file),
-            "--text", text,
-            "--length_scale", str(length_scale),
-            "--noise_scale", str(noise_scale),
-            "--noise_w", str(noise_w),
-        ]
-
-        result = subprocess.run(command, capture_output=True, text=True)
-        if result.returncode != 0 or not output_file.exists():
-            raise HTTPException(status_code=500, detail=f"TTS failed: {result.stderr.strip()}")
-
-        return {
-            "ok": True,
-            "voice": voice,
-            "audio_url": f"/file/{output_file.name}"
-        }
+        while True:
+            data = await websocket.receive_text()
+
+            if data.startswith("{") and "text" in data:
+                import json
+                payload = json.loads(data)
+                text = payload.get("text", "")
+                voice = payload.get("voice", DEFAULT_VOICE)
+                settings["length_scale"] = float(payload.get("length_scale", 1.0))
+                settings["noise_scale"] = float(payload.get("noise_scale", 0.33))
+                settings["noise_w"] = float(payload.get("noise_w", 0.5))
+
+                tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
+                cmd = [
+                    "piper",
+                    "--model", str(VOICES_DIR / f"{voice}.onnx"),
+                    "--output_file", tmp.name,
+                    "--text", text,
+                    "--length_scale", str(settings["length_scale"]),
+                    "--noise_scale", str(settings["noise_scale"]),
+                    "--noise_w", str(settings["noise_w"]),
+                ]
+                subprocess.run(cmd, check=True)
+
+                with open(tmp.name, "rb") as f:
+                    audio = f.read()
+                    await websocket.send_bytes(audio)
+
+                tmp.close()
+                os.unlink(tmp.name)
 
     except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-@app.get("/file/{filename}")
-async def get_file(filename: str):
-    path = FILES_DIR / filename
-    if path.exists():
-        return FileResponse(path)
-    raise HTTPException(status_code=404, detail="File not found")
+        print(f"[TTS WS Error] {e}")
+    finally:
+        wave_writer.close()
+        await websocket.close()