Fix M4T for ASR pipeline (#32296)

* tentative fix

* do the same for M4T

src/transformers/models/seamless_m4t/modeling_seamless_m4t.py

@@ -3154,6 +3154,7 @@ class SeamlessM4TForSpeechToText(SeamlessM4TPreTrainedModel):
         """
         text_decoder_input_ids = kwargs.pop("decoder_input_ids", None)
         # overwrite text_decoder_input_ids if tgt_lang is passed. The latter gets priority over decoder_input_ids.
+        input_features = input_features if input_features is not None else kwargs.pop("inputs")
         if tgt_lang is not None:
             inputs = kwargs.get("input_embeds") if input_features is None else input_features
             inputs = (

src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py

@@ -3422,6 +3422,7 @@ class SeamlessM4Tv2ForSpeechToText(SeamlessM4Tv2PreTrainedModel):
         """
         text_decoder_input_ids = kwargs.pop("decoder_input_ids", None)
         # overwrite text_decoder_input_ids if tgt_lang is passed. The latter gets priority over decoder_input_ids.
+        input_features = input_features if input_features is not None else kwargs.pop("inputs")
         if tgt_lang is not None:
             inputs = kwargs.get("input_embeds") if input_features is None else input_features
             inputs = (