diff --git a/api/core/model_runtime/model_providers/xinference/xinference_helper.py b/api/core/model_runtime/model_providers/xinference/xinference_helper.py
index 75161ad376c419..151166f165cb70 100644
--- a/api/core/model_runtime/model_providers/xinference/xinference_helper.py
+++ b/api/core/model_runtime/model_providers/xinference/xinference_helper.py
@@ -89,7 +89,7 @@ def _get_xinference_extra_parameter(server_url: str, model_uid: str, api_key: st
             model_handle_type = 'embedding'
         elif response_json.get('model_type') == 'audio':
             model_handle_type = 'audio'
-            if model_family and model_family in ['ChatTTS', 'CosyVoice']:
+            if model_family and model_family in ['ChatTTS', 'CosyVoice', 'FishAudio']:
                 model_ability.append('text-to-audio')
             else:
                 model_ability.append('audio-to-text')