Only modify cues for auto-generated captions

src/invidious.cr

@@ -3203,7 +3203,12 @@ get "/api/v1/captions/:id" do |env|
     caption = caption[0]
   end
 
-  caption_xml = client.get(caption.baseUrl + "&tlang=#{tlang}").body
+  url = caption.baseUrl + "&tlang=#{tlang}"
+
+  # Auto-generated captions often have cues that aren't aligned properly with the video,
+  # as well as some other markup that makes it cumbersome, so we try to fix that here
+  if caption.name.simpleText.includes? "auto-generated"
+    caption_xml = client.get(url).body
     caption_xml = XML.parse(caption_xml)
 
     webvtt = <<-END_VTT
@@ -3243,6 +3248,10 @@ get "/api/v1/captions/:id" do |env|
 
     END_CUE
     end
+  else
+    url += "&format=vtt"
+    webvtt = client.get(url).body
+  end
 
   if title = env.params.query["title"]?
     # https://blog.fastmail.com/2011/06/24/download-non-english-filenames/