Add torchcodec in docstrings/tests for datasets 4.0 (#39156)

* fix dataset run_object_detection

* bump version

* keep same dataset actually

* torchcodec in docstrings and testing utils

* torchcodec in dockerfiles and requirements

* remove duplicate

* add torchocodec to all the remaining docker files

* fix tests

* support torchcodec in audio classification and ASR

* [commit to revert] build ci-dev images

* [commit to revert] trigger circleci

* [commit to revert] build ci-dev images

* fix

* fix modeling_hubert

* backward compatible run_object_detection

* revert ci trigger commits

* fix mono conversion and support torch tensor as input

* revert map_to_array docs + fix it

* revert mono

* nit in docstring

* style

* fix modular

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

tests/models/phi4_multimodal/test_feature_extractor_phi4_multimodal.py

@@ -207,7 +207,7 @@ class Phi4MultimodalFeatureExtractionTest(SequenceFeatureExtractionTestMixin, un
     def _load_datasamples(self, num_samples):
         ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
         # automatic decoding with librispeech
-        speech_samples = ds.sort("id").select(range(num_samples))[:num_samples]["audio"]
+        speech_samples = ds.sort("id")[:num_samples]["audio"]
 
         return [x["array"] for x in speech_samples]
 

tests/models/phi4_multimodal/test_modeling_phi4_multimodal.py

@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import tempfile
 import unittest
 
 import requests
@@ -33,13 +32,13 @@ from transformers import (
 from transformers.testing_utils import (
     Expectations,
     cleanup,
-    require_soundfile,
     require_torch,
     require_torch_large_accelerator,
+    require_torchcodec,
     slow,
     torch_device,
 )
-from transformers.utils import is_soundfile_available
+from transformers.utils import is_torchcodec_available
 
 from ...generation.test_utils import GenerationTesterMixin
 from ...test_configuration_common import ConfigTester
@@ -54,8 +53,8 @@ if is_vision_available():
     from PIL import Image
 
 
-if is_soundfile_available():
-    import soundfile
+if is_torchcodec_available():
+    import torchcodec
 
 
 class Phi4MultimodalModelTester:
@@ -296,11 +295,9 @@ class Phi4MultimodalIntegrationTest(unittest.TestCase):
         self.assistant_token = "<|assistant|>"
         self.end_token = "<|end|>"
         self.image = Image.open(requests.get(self.image_url, stream=True).raw)
-        with tempfile.NamedTemporaryFile(mode="w+b", suffix=".wav") as tmp:
-            tmp.write(requests.get(self.audio_url, stream=True).raw.data)
-            tmp.flush()
-            tmp.seek(0)
-            self.audio, self.sampling_rate = soundfile.read(tmp.name)
+        audio_bytes = requests.get(self.audio_url, stream=True).raw.data
+        samples = torchcodec.decoders.AudioDecoder(audio_bytes).get_all_samples()
+        self.audio, self.sampling_rate = samples.data, samples.sample_rate
 
         cleanup(torch_device, gc_collect=True)
 
@@ -378,7 +375,7 @@ class Phi4MultimodalIntegrationTest(unittest.TestCase):
 
         self.assertEqual(response, EXPECTED_RESPONSE)
 
-    @require_soundfile
+    @require_torchcodec
     def test_audio_text_generation(self):
         model = AutoModelForCausalLM.from_pretrained(
             self.checkpoint_path, revision=self.revision, torch_dtype=torch.float16, device_map=torch_device