Add custom inference handler for Maya1 TTS

handler.py

@@ -274,7 +274,7 @@ class EndpointHandler:
             wav_bytes = buf.getvalue()
             b64 = base64.b64encode(wav_bytes).decode("utf-8")
 
-            return {"audio_base64": b64, "sampling_rate": sr}
+            return [{"blob": b64, "content-type": "audio/wav"}]
 
         # Build properly formatted prompt for Maya1
         prompt = self.build_prompt(description, text)
@@ -337,10 +337,9 @@ class EndpointHandler:
         wav_bytes = buf.getvalue()
         b64 = base64.b64encode(wav_bytes).decode("utf-8")
 
-        return {
-            "audio_base64": b64,
-            "sampling_rate": 24000,
-        }
+        # Return in HuggingFace Inference Endpoints format
+        # The endpoint expects a list with blob data
+        return [{"blob": b64, "content-type": "audio/wav"}]
 # Module-level convenience functions for hosting platforms (Hugging Face Endpoints)
 # The platform typically expects top-level `init` and `predict` (or `run`) callables
 # so we provide thin wrappers around the EndpointHandler class.