Add generate kwargs to `AutomaticSpeechRecognitionPipeline` (#20952)

* Add generate kwargs to AutomaticSpeechRecognitionPipeline

* Add test for generation kwargs

src/transformers/pipelines/automatic_speech_recognition.py

@@ -24,6 +24,8 @@ from .base import ChunkPipeline
 
 
 if TYPE_CHECKING:
+    from pyctcdecode import BeamSearchDecoderCTC
+
     from ...feature_extraction_sequence_utils import SequenceFeatureExtractor
 
 logger = logging.get_logger(__name__)
@@ -169,8 +171,14 @@ class AutomaticSpeechRecognitionPipeline(ChunkPipeline):
 
     """
 
-    def __init__(self, feature_extractor: Union["SequenceFeatureExtractor", str], *args, **kwargs):
-        super().__init__(*args, **kwargs)
+    def __init__(
+        self,
+        feature_extractor: Union["SequenceFeatureExtractor", str],
+        *,
+        decoder: Optional[Union["BeamSearchDecoderCTC", str]] = None,
+        **kwargs
+    ):
+        super().__init__(**kwargs)
         self.feature_extractor = feature_extractor
 
         if self.model.__class__ in MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING.values():
@@ -178,9 +186,9 @@ class AutomaticSpeechRecognitionPipeline(ChunkPipeline):
         elif (
             feature_extractor._processor_class
             and feature_extractor._processor_class.endswith("WithLM")
-            and kwargs.get("decoder", None) is not None
+            and decoder is not None
         ):
-            self.decoder = kwargs["decoder"]
+            self.decoder = decoder
             self.type = "ctc_with_lm"
         else:
             self.type = "ctc"
@@ -221,6 +229,12 @@ class AutomaticSpeechRecognitionPipeline(ChunkPipeline):
                 `timestamps` along the text for every word in the text. For instance if you get `[{"text": "hi ",
                 "timestamps": (0.5,0.9), {"text": "there", "timestamps": (1.0, .1.5)}]`, then it means the model
                 predicts that the word "hi" was pronounced after `0.5` and before `0.9` seconds.
+            generate_kwargs (`dict`, *optional*):
+                The dictionary of ad-hoc parametrization of `generate_config` to be used for the generation call. For a
+                complete overview of generate, check the [following
+                guide](https://huggingface.co/docs/transformers/en/main_classes/text_generation).
+            max_new_tokens (`int`, *optional*):
+                The maximum numbers of tokens to generate, ignoring the number of tokens in the prompt.
 
         Return:
             `Dict`: A dictionary with the following keys:
@@ -233,23 +247,43 @@ class AutomaticSpeechRecognitionPipeline(ChunkPipeline):
         """
         return super().__call__(inputs, **kwargs)
 
-    def _sanitize_parameters(self, **kwargs):
+    def _sanitize_parameters(
+        self,
+        chunk_length_s=None,
+        stride_length_s=None,
+        ignore_warning=None,
+        decoder_kwargs=None,
+        return_timestamps=None,
+        generate_kwargs=None,
+        max_new_tokens=None,
+    ):
         # No parameters on this pipeline right now
         preprocess_params = {}
-        if "chunk_length_s" in kwargs:
-            preprocess_params["chunk_length_s"] = kwargs["chunk_length_s"]
-        if "stride_length_s" in kwargs:
-            preprocess_params["stride_length_s"] = kwargs["stride_length_s"]
-        if "ignore_warning" in kwargs:
-            preprocess_params["ignore_warning"] = kwargs["ignore_warning"]
+        if chunk_length_s is not None:
+            preprocess_params["chunk_length_s"] = chunk_length_s
+        if stride_length_s is not None:
+            preprocess_params["stride_length_s"] = stride_length_s
+        if ignore_warning is not None:
+            preprocess_params["ignore_warning"] = ignore_warning
+
+        forward_params = {"generate_kwargs": {}}
+        if max_new_tokens is not None:
+            forward_params["generate_kwargs"]["max_new_tokens"] = max_new_tokens
+        if generate_kwargs is not None:
+            if max_new_tokens is not None and "max_new_tokens" in generate_kwargs:
+                raise ValueError(
+                    "`max_new_tokens` is defined both as an argument and inside `generate_kwargs` argument, please use"
+                    " only 1 version"
+                )
+            forward_params["generate_kwargs"].update(generate_kwargs)
 
         postprocess_params = {}
-        if "decoder_kwargs" in kwargs:
-            postprocess_params["decoder_kwargs"] = kwargs["decoder_kwargs"]
-        if "return_timestamps" in kwargs:
-            postprocess_params["return_timestamps"] = kwargs["return_timestamps"]
+        if decoder_kwargs is not None:
+            postprocess_params["decoder_kwargs"] = decoder_kwargs
+        if return_timestamps is not None:
+            postprocess_params["return_timestamps"] = return_timestamps
 
-        return preprocess_params, {}, postprocess_params
+        return preprocess_params, forward_params, postprocess_params
 
     def preprocess(self, inputs, chunk_length_s=0, stride_length_s=None, ignore_warning=False):
         if isinstance(inputs, str):
@@ -351,7 +385,10 @@ class AutomaticSpeechRecognitionPipeline(ChunkPipeline):
                 processed["stride"] = stride
             yield {"is_last": True, **processed, **extra}
 
-    def _forward(self, model_inputs):
+    def _forward(self, model_inputs, generate_kwargs=None):
+        if generate_kwargs is None:
+            generate_kwargs = {}
+
         is_last = model_inputs.pop("is_last")
         if self.type == "seq2seq":
             encoder = self.model.get_encoder()
@@ -376,6 +413,7 @@ class AutomaticSpeechRecognitionPipeline(ChunkPipeline):
             tokens = self.model.generate(
                 encoder_outputs=encoder(inputs, attention_mask=attention_mask),
                 attention_mask=attention_mask,
+                **generate_kwargs,
             )
 
             out = {"tokens": tokens}

tests/pipelines/test_pipelines_automatic_speech_recognition.py

@@ -169,6 +169,17 @@ class AutomaticSpeechRecognitionPipelineTests(unittest.TestCase, metaclass=Pipel
         output = speech_recognizer(waveform)
         self.assertEqual(output, {"text": "あл ش 湯 清 ه ܬ া लᆨしث ल eか u w 全 u"})
 
+    @require_torch
+    def test_small_model_pt_seq2seq_gen_kwargs(self):
+        speech_recognizer = pipeline(
+            model="hf-internal-testing/tiny-random-speech-encoder-decoder",
+            framework="pt",
+        )
+
+        waveform = np.tile(np.arange(1000, dtype=np.float32), 34)
+        output = speech_recognizer(waveform, max_new_tokens=10, generate_kwargs={"num_beams": 2})
+        self.assertEqual(output, {"text": "あл † γ ت ב オ 束 泣 足"})
+
     @slow
     @require_torch
     @require_pyctcdecode