From 8929a1ce7bd8afb9da6a34443b71c2816c40c680 Mon Sep 17 00:00:00 2001
From: Sacha Chua <sacha@sachachua.com>
Date: Sun, 23 Oct 2022 08:32:04 -0400
Subject: Captioning tools

---
 roles/caption/tasks/main.yml       |  6 ++++++
 roles/caption/templates/caption.sh | 18 ++++++++++++++++++
 2 files changed, 24 insertions(+)
 create mode 100755 roles/caption/templates/caption.sh

diff --git a/roles/caption/tasks/main.yml b/roles/caption/tasks/main.yml
index 7b6534a..7bf45d0 100644
--- a/roles/caption/tasks/main.yml
+++ b/roles/caption/tasks/main.yml
@@ -6,3 +6,9 @@
 - name: Install whisper
   ansible.builtin.pip:
     name: git+https://github.com/openai/whisper.git
+- name: Copy the shell script
+  tags: caption-sh
+  template:
+    src: caption.sh
+    dest: current
+    mode: 0755
diff --git a/roles/caption/templates/caption.sh b/roles/caption/templates/caption.sh
new file mode 100755
index 0000000..9600a3c
--- /dev/null
+++ b/roles/caption/templates/caption.sh
@@ -0,0 +1,18 @@
+#!/bin/bash
+# {{ ansible_managed }}
+FILE="$1"
+MODEL="${2:small}"
+AUDIO=$(basename "$FILE" | sed s/\\.[a-z][a-z][a-z][a-z]?$//).ogg
+if [[ ! -f $AUDIO ]]; then
+  if  [[ "$FILE" == *webm ]]; then
+      ffmpeg -y -i "$FILE" -acodec copy -vn $AUDIO
+  else
+      ffmpeg -y -i "$FILE" -acodec libvorbis -vn $AUDIO
+  fi
+fi
+date > $AUDIO-$MODEL.log
+time whisper $AUDIO --model $MODEL --threads 12 >> $AUDIO-$MODEL.log
+for EXT in vtt txt srt; do
+    mv $AUDIO.$EXT $(basename -s .webm.$EXT $AUDIO.$EXT)
+done
+date >> $AUDIO-$MODEL.log
-- 
cgit v1.2.3