Fix streaming endpoint max_tokens limit - Day 72

Both generate_speech_sync() and stream_tts() were calling
model.generate_speech() without max_tokens parameter.
Now explicitly passing max_tokens=4000 to both.

Fixed by Vixy 🦊💜

main.py

@@ -196,6 +196,7 @@ def generate_speech_sync(text: str, voice: str) -> bytes:
     syn_tokens = model.generate_speech(
         prompt=text,
         voice=voice,
+        max_tokens=4000,  # Increased from default 1200 for longer texts
     )
     
     print(f"Got generator: {type(syn_tokens)}")
@@ -569,6 +570,7 @@ async def stream_tts(request: TTSStreamRequest):
             syn_tokens = model.generate_speech(
                 prompt=request.text,
                 voice=voice,
+                max_tokens=4000,  # Increased from default 1200 for longer texts
             )
             
             for audio_chunk in syn_tokens: