diff --git a/tests/pipelines/test_pipelines_text_to_audio.py b/tests/pipelines/test_pipelines_text_to_audio.py
index e07e2ad392a3..dac2ce6b30ec 100644
--- a/tests/pipelines/test_pipelines_text_to_audio.py
+++ b/tests/pipelines/test_pipelines_text_to_audio.py
@@ -27,6 +27,7 @@
     require_torch,
     require_torch_accelerator,
     require_torch_or_tf,
+    run_test_using_subprocess,
     slow,
     torch_device,
 )
@@ -66,8 +67,10 @@ def test_small_musicgen_pt(self):
         audio = [output["audio"] for output in outputs]
         self.assertEqual([ANY(np.ndarray), ANY(np.ndarray)], audio)
 
+    # TODO: @ylacombe: `SeamlessM4TForTextToSpeech.generate` has issue with `generation_config`. See issue #34811
     @slow
     @require_torch
+    @run_test_using_subprocess
     def test_medium_seamless_m4t_pt(self):
         speech_generator = pipeline(task="text-to-audio", model="facebook/hf-seamless-m4t-medium", framework="pt")