fix: ZONOS2 only generating 1024 tokens max

1 files changed, 6 insertions, 0 deletions

diff --git a/app/converter/clients/sglomni.py b/app/converter/clients/sglomni.py
index 8484aba..d4b756f 100644
--- a/app/converter/clients/sglomni.py
+++ b/app/converter/clients/sglomni.py
@@ -274,6 +274,12 @@ class SgOmniTTSClient(BaseTTSClient):
         }
         if self._seed is not None:
             payload["seed"] = self._seed
+        if entry.max_new_tokens is not None:
+            # Models whose engine caps a request below what a full
+            # sub-chunk can narrate (Zonos2's 1024-frame default is ~12 s):
+            # raise the ceiling per request. Generation still stops at
+            # natural EOS, so an unused margin costs nothing.
+            payload["max_new_tokens"] = entry.max_new_tokens
         if entry.capability == "design":
             payload["task_type"] = "VoiceDesign"
             payload["instructions"] = self.instructions