Add script to download whisperx models

guardian · Jan 21, 2025 · f7b8e7b · f7b8e7b
1 parent 66090cd
commit f7b8e7b
Showing 1 changed file with 23 additions and 0 deletions.
diff --git a/roles/whisperx/tasks/main.yml b/roles/whisperx/tasks/main.yml
@@ -19,3 +19,26 @@
     name: whisperx
     executable: pip3
 
+- name: Install dependencies needed for downloading models
+  pip:
+    name:
+     - huggingface_hub
+     - typer
+    executable: pip3
+
+- name: Download model fetch script
+  get_url:
+      url: "https://raw.githubusercontent.com/guardian/transcription-service/refs/heads/add-whisperx-support/scripts/download_whisperx_models.py"
+      dest: "/tmp/download_whisperx_models.py"
+      mode: "0755"
+
+- name: Download whisperx models
+  command: "python3 /tmp/download_whisperx_models.py --whisper-models --diarization-models --torch-align-models --huggingface-token {{ huggingface_token }}"
+
+- name: Remove download script dependencies
+  pip:
+    name:
+     - huggingface_hub
+     - typer
+    state: absent
+    executable: pip3