Aligned with tts

processing_moss_tts.py

@@ -159,7 +159,7 @@ class UserMessage(Message):
             reference = []
             for speaker_idx, speaker_reference in enumerate(self.reference):
                 if speaker_reference is not None:
-                    reference.append(f"[S{speaker_idx}]:\n{AUDIO_PLACEHOLDER}")
+                    reference.append(f"[S{speaker_idx+1}]:\n{AUDIO_PLACEHOLDER}")
             reference = "\n".join(reference)
             audio_codes_list = [
                 speaker_reference
@@ -631,6 +631,7 @@ class MossTTSDelayProcessor(ProcessorMixin):
         """
         if role == "user":
             audio_gen_slot_token = audio_delay_slot_token = self.audio_user_slot_token
+            truncation = False
         else:
             audio_gen_slot_token = self.audio_assistant_gen_slot_token
             audio_delay_slot_token = self.audio_assistant_delay_slot_token