[pipeline] ASR pipeline kwargs are forwared to generate

src/transformers/models/whisper/generation_whisper.py

@@ -1594,7 +1594,7 @@ def language_to_id(language: str) -> int:
                     # if task is defined it'll overwrite task ids that might have already been defined via the generation_config
                     replace_or_add(init_tokens[i], task_id, generation_config.task_to_id.values())
                 else:
-                    raise ValueError(f"The `{task}`task is not supported. The task should be one of `{TASK_IDS}`")
+                    raise ValueError(f"The `{task}` task is not supported. The task should be one of `{TASK_IDS}`")
             elif language is not None and hasattr(generation_config, "task_to_id"):
                 # if language is defined, but no task id is in `init_tokens`, default to transcribe
                 if not any(ti in init_tokens[i] for ti in generation_config.task_to_id.values()):

src/transformers/pipelines/automatic_speech_recognition.py

@@ -282,10 +282,13 @@ def _sanitize_parameters(
         decoder_kwargs=None,
         return_timestamps=None,
         return_language=None,
-        generate_kwargs=None,
+        **generate_kwargs,
     ):
-        # No parameters on this pipeline right now
         preprocess_params = {}
+        forward_params = {}
+        postprocess_params = {}
+
+        # Preprocess params
         if chunk_length_s is not None:
             if self.type in ["seq2seq", "seq2seq_whisper"] and not ignore_warning:
                 type_warning = (
@@ -305,14 +308,28 @@ def _sanitize_parameters(
         if stride_length_s is not None:
             preprocess_params["stride_length_s"] = stride_length_s
 
-        forward_params = defaultdict(dict)
-        if generate_kwargs is not None:
-            forward_params.update(generate_kwargs)
+        # Forward params
+        # BC: accept a dictionary of generation kwargs (as opposed to **generate_kwargs)
+        if "generate_kwargs" in generate_kwargs:
+            forward_params.update(generate_kwargs.pop("generate_kwargs"))
+        # Default use for kwargs: they are generation-time kwargs
+        forward_params.update(generate_kwargs)
 
-        postprocess_params = {}
+        if getattr(self, "assistant_model", None) is not None:
+            forward_params["assistant_model"] = self.assistant_model
+        if getattr(self, "assistant_tokenizer", None) is not None:
+            forward_params["tokenizer"] = self.tokenizer
+            forward_params["assistant_tokenizer"] = self.assistant_tokenizer
+
+        # Postprocess params
         if decoder_kwargs is not None:
             postprocess_params["decoder_kwargs"] = decoder_kwargs
+        if return_language is not None:
+            if self.type != "seq2seq_whisper":
+                raise ValueError("Only Whisper can return language for now.")
+            postprocess_params["return_language"] = return_language
 
+        # Parameter used in more than one place
         # in some models like whisper, the generation config has a `return_timestamps` key
         if hasattr(self, "generation_config") and hasattr(self.generation_config, "return_timestamps"):
             return_timestamps = return_timestamps or self.generation_config.return_timestamps
@@ -335,16 +352,6 @@ def _sanitize_parameters(
                 )
             forward_params["return_timestamps"] = return_timestamps
             postprocess_params["return_timestamps"] = return_timestamps
-        if return_language is not None:
-            if self.type != "seq2seq_whisper":
-                raise ValueError("Only Whisper can return language for now.")
-            postprocess_params["return_language"] = return_language
-
-        if getattr(self, "assistant_model", None) is not None:
-            forward_params["assistant_model"] = self.assistant_model
-        if getattr(self, "assistant_tokenizer", None) is not None:
-            forward_params["tokenizer"] = self.tokenizer
-            forward_params["assistant_tokenizer"] = self.assistant_tokenizer
 
         return preprocess_params, forward_params, postprocess_params
 

tests/pipelines/test_pipelines_automatic_speech_recognition.py

@@ -1791,6 +1791,32 @@ def test_pipeline_assisted_generation(self):
         with self.assertRaises(ValueError):
             _ = pipe(prompt, generate_kwargs={"num_beams": 2})
 
+    @require_torch
+    def test_pipeline_generation_kwargs(self):
+        """Tests that we can pass kwargs to `generate`, as in the text generation pipelines"""
+        model = "openai/whisper-tiny"
+        asr = pipeline("automatic-speech-recognition", model=model)
+        dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation[:1]")
+
+        # BC: with `generate_kwargs` as a dictionary
+        res = asr(
+            dataset[0]["audio"],
+            generate_kwargs={"task": "transcribe", "max_new_tokens": 256},
+        )
+        self.assertEqual(
+            res["text"], " Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel."
+        )
+
+        # New: kwargs forwarded to `generate`
+        res = asr(
+            dataset[0]["audio"],
+            max_new_tokens=256,
+            task="transcribe",
+        )
+        self.assertEqual(
+            res["text"], " Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel."
+        )
+
 
 def require_ffmpeg(test_case):
     """