added proper conversion

main.py

@@ -157,8 +157,18 @@ def _synthesize_to_bytes(text: str, voice: str, speed: float, output_format: str
         raise RuntimeError("No audio generated")
     final_audio = np.concatenate(all_audio)
     duration = len(final_audio) / 24000
+    
     buf = io.BytesIO()
     sf.write(buf, final_audio, 24000, format="WAV")
+    
+    if output_format == "mp3":
+        buf.seek(0)
+        from pydub import AudioSegment
+        audio_segment = AudioSegment.from_wav(buf)
+        mp3_buf = io.BytesIO()
+        audio_segment.export(mp3_buf, format="mp3")
+        return mp3_buf.getvalue(), duration
+        
     return buf.getvalue(), duration
 
 # ── Routes ─────────────────────────────────────────────────────────────────────