Update app.py

app.py

@@ -1,44 +1,119 @@
-import gradio as gr
-from gradio import Server
-from transformers import T5ForConditionalGeneration, T5Tokenizer
+import requests
+import urllib.parse
+from typing import Optional, Dict, Any
 
-# Load model and tokenizer
-MODEL_NAME = "alirezamsh/small100"
-print("Loading model...")
-tokenizer = T5Tokenizer.from_pretrained(MODEL_NAME)
-model = T5ForConditionalGeneration.from_pretrained(MODEL_NAME)
-print("Model loaded!")
+app = Server() 
 
 
-app = Server() 
 
-@app.api(name="translate")
-def translate(text, prompt):
+def generate_tts_audio(
+    text: str,
+    voice: str = "Sam",
+    pitch: int = 100,
+    speed: int = 100,
+    base_url: str = "https://tetyys.com/SAPI4"
+) -> bytes:
     """
-    Translate text using T5 with a task prefix.
-    Note: t5-small is trained mainly on English tasks and has limited
-    multilingual translation ability. See note at the bottom.
+    Generate TTS audio using the Tetyys SAPI4 web interface.
+    
+    Args:
+        text: Text to synthesize (required)
+        voice: Voice name (optional, default: "Sam")
+        pitch: Pitch value (optional, default: 100)
+        speed: Speed value (optional, default: 100)
+        base_url: Base URL for the API (optional)
+    
+    Returns:
+        bytes: WAV audio file content
+    
+    Raises:
+        requests.RequestException: If the HTTP request fails
+        ValueError: If the response is not valid audio
     """
-    if not text.strip():
-        return "Please enter some text to translate."
+    # URL encode the text to handle special characters
+    encoded_text = urllib.parse.quote(text)
+    
+    # Build the endpoint URL with query parameters
+    endpoint = f"{base_url}/SAPI4?text={encoded_text}"
+    
+    # Add optional parameters if provided
+    params = []
+    if voice:
+        params.append(f"voice={urllib.parse.quote(voice)}")
+    if pitch is not None:
+        params.append(f"pitch={pitch}")
+    if speed is not None:
+        params.append(f"speed={speed}")
+    
+    # Combine base endpoint with optional parameters
+    if params:
+        endpoint += "&" + "&".join(params)
+    
+    # Make the request
+    response = requests.get(endpoint, timeout=30)
+    response.raise_for_status()
+    
+    # Validate that we received audio content
+    content_type = response.headers.get('Content-Type', '')
+    if 'audio' not in content_type.lower():
+        raise ValueError(
+            f"Expected audio response, got: {content_type}. "
+            f"Response preview: {response.text[:100]}"
+        )
+    
+    return response.content
 
-    tokenizer(prompt, return_tensors="pt", max_length=512, truncation=True)
 
-    with torch.no_grad():
-        outputs = model.generate(
-            **inputs,
-            max_length=512,
-            num_beams=4,
-            early_stopping=True,
-        )
 
-    result = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    return result
+@app.api(name="tts")
+def save_tts_audio(
+    text: str,
+    output_path: str,
+    voice: str = "Sam",
+    pitch: int = 100,
+    speed: int = 100,
+    base_url: str = "https://tetyys.com/SAPI4"
+) -> None:
+    """
+    Generate TTS audio and save it to a file.
+    
+    Args:
+        text: Text to synthesize
+        output_path: Path where the WAV file will be saved
+        voice: Voice name (optional, default: "Sam")
+        pitch: Pitch value (optional, default: 100)
+        speed: Speed value (optional, default: 100)
+        base_url: Base URL for the API (optional)
+    """
+    audio_data = generate_tts_audio(text, voice, pitch, speed, base_url)
+    
+    with open(output_path, 'wb') as f:
+        f.write(audio_data)
+
 
+def get_voice_limits(
+    voice: str,
+    base_url: str = "https://tetyys.com/SAPI4"
+) -> Dict[str, Any]:
+    """
+    Get the pitch and speed limitations for a specific voice.
+    
+    Args:
+        voice: Voice name to query
+        base_url: Base URL for the API (optional)
+    
+    Returns:
+        dict: Voice limitations data
+    """
+    encoded_voice = urllib.parse.quote(voice)
+    endpoint = f"{base_url}/VoiceLimitations?voice={encoded_voice}"
+    
+    response = requests.get(endpoint, timeout=30)
+    response.raise_for_status()
+    
+    return response.json()
 
 
 
-# ─────────────────────────────────────────────
-# 5. LAUNCH
 # ─────────────────────────────────────────────
 app.launch()