Update app.py

app.py

@@ -3,6 +3,7 @@ from fastapi.responses import StreamingResponse
 from piper import PiperVoice
 import io
 import os
+from pydantic import BaseModel
 
 app = FastAPI()
 
@@ -10,6 +11,34 @@ app = FastAPI()
 MODEL_DIR = "./models"
 os.makedirs(MODEL_DIR, exist_ok=True)
 
+VOICE_MAP = {
+    # Gendered Languages (Male and Female models available)
+    "en": {"gendered": True, "male": "en_GB-alan-medium", "female": "en_GB-semaine-medium"},
+    "es": {"gendered": True, "male": "es_ES-sharvard-medium", "female": "es_ES-carlota-low"},
+    "fr": {"gendered": True, "male": "fr_FR-upmc-medium", "female": "fr_FR-siwis-low"},
+    "de": {"gendered": True, "male": "de_DE-thorsten-medium", "female": "de_DE-kerstin-low"},
+    "it": {"gendered": True, "male": "it_IT-riccardo-x_low", "female": "it_IT-paola-medium"},
+    "pl": {"gendered": True, "male": "pl_PL-dark_ness-medium", "female": "pl_PL-gosia-medium"},
+    "uk": {"gendered": True, "male": "uk_UA-ukrainian_tts-medium", "female": "uk_UA-lada-medium"},
+    "nl": {"gendered": True, "male": "nl_NL-ronnie-medium", "female": "nl_NL-mls-medium"},
+
+    # Non-Gendered / Single-Voice Languages (Default model used)
+    "bg": {"gendered": False, "default": "bg_BG-dimitar-medium"},
+    "ca": {"gendered": False, "default": "ca_ES-upc_ona-medium"},
+    "cs": {"gendered": False, "default": "cs_CZ-jirka-medium"},
+    "da": {"gendered": False, "default": "da_DK-talesyntese-medium"},
+    "fi": {"gendered": False, "default": "fi_FI-harri-medium"},
+    "el": {"gendered": False, "default": "el_GR-rapotakis-low"},
+    "hu": {"gendered": False, "default": "hu_HU-mls-medium"},
+    "is": {"gendered": False, "default": "is_IS-ugla-medium"},
+    "lv": {"gendered": False, "default": "lv_LV-sanda-medium"},
+    "ro": {"gendered": False, "default": "ro_RO-mls-medium"},
+    "sk": {"gendered": False, "default": "sk_SK-lili-medium"},
+    "sl": {"gendered": False, "default": "sl_SI-artificer-medium"},
+    "sv": {"gendered": False, "default": "sv_SE-extf0-medium"},
+    "cy": {"gendered": False, "default": "cy_GB-bu-tts-medium"}
+}
+
 # Cache for loaded models to avoid re-loading from disk every request
 loaded_voices = {}
 
@@ -25,20 +54,42 @@ def get_voice(model_name: str):
         loaded_voices[model_name] = PiperVoice.load(model_path, config_path)
     return loaded_voices[model_name]
 
-@app.get("/tts")
-async def tts(text: str, model: str = "en_US-lessac-medium"):
+class TTSRequest(BaseModel):
+    text: str
+    language: str
+    gender: str
+
+@app.post("/tts")
+async def tts_post(request: TTSRequest):
     try:
-        voice = get_voice(model)
-        
-        # Create an in-memory buffer for the WAV file
+        lang_code = request.language.lower()
+        gender_req = request.gender.lower()
+
+        # 1. Get the language entry
+        lang_entry = VOICE_MAP.get(lang_code)
+        if not lang_entry:
+            raise HTTPException(status_code=400, detail=f"Language '{lang_code}' not supported.")
+
+        # 2. Determine which model file to use
+        if lang_entry["gendered"]:
+            # If gendered, try to get the requested gender
+            model_name = lang_entry.get(gender_req)
+            if not model_name:
+                # Fallback to male if requested gender isn't found in a gendered set
+                model_name = lang_entry["male"] 
+        else:
+            # If not gendered, use the default key
+            model_name = lang_entry["default"]
+
+        # 3. Load and Synthesize
+        voice = get_voice(model_name)
         wav_buffer = io.BytesIO()
-        with io.BytesIO() as f:
-            voice.synthesize(text, f)
-            audio_data = f.getvalue()
-            
-        return Response(content=audio_data, media_type="audio/wav")
+        voice.synthesize(request.text, wav_buffer)
+        
+        return Response(content=wav_buffer.getvalue(), media_type="audio/wav")
+
     except Exception as e:
-        return {"error": str(e)}
+        raise HTTPException(status_code=500, detail=str(e))
 
 @app.get("/health")
 def home():
@@ -46,4 +97,4 @@ def home():
 
 @app.get("/")
 def home():
-    return {"status": "Piper TTS is running"}
+    return {"languages": VOICE_MAP}