diff --git a/psifx/audio/diarization/pyannote/command.py b/psifx/audio/diarization/pyannote/command.py
index 3707446..dd4cf5e 100644
--- a/psifx/audio/diarization/pyannote/command.py
+++ b/psifx/audio/diarization/pyannote/command.py
@@ -53,8 +53,8 @@ def setup(parser: argparse.ArgumentParser):
         parser.add_argument(
             "--model_name",
             type=str,
-            default="2.1.1",
-            help="version number of the pyannote/speaker-diarization model, c.f."
+            default="pyannote/speaker-diarization@2.1.1",
+            help="name of the diarization model used, c.f."
             " https://huggingface.co/pyannote/speaker-diarization/tree/main/reproducible_research",
         )
         parser.add_argument(
diff --git a/psifx/audio/diarization/pyannote/tool.py b/psifx/audio/diarization/pyannote/tool.py
index 3dabde7..9d07a7e 100644
--- a/psifx/audio/diarization/pyannote/tool.py
+++ b/psifx/audio/diarization/pyannote/tool.py
@@ -18,7 +18,7 @@ class PyannoteDiarizationTool(DiarizationTool):
 
     def __init__(
         self,
-        model_name: str = "2.1.1",
+        model_name: str = "pyannote/speaker-diarization@2.1.1",
         api_token: Optional[str] = None,
         device: str = "cpu",
         overwrite: bool = False,
@@ -34,7 +34,7 @@ def __init__(
         self.api_token = api_token
 
         self.model: Pipeline = Pipeline.from_pretrained(
-            checkpoint_path=f"pyannote/speaker-diarization@{model_name}",
+            checkpoint_path=model_name,
             use_auth_token=api_token,
         ).to(device=torch.device(device))