Create VTT files from HLS derivative

lehigh-university-libraries · Dec 16, 2024 · daa9041 · daa9041
1 parent 2b04619
commit daa9041
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 5 deletions.
diff --git a/examples/whisper/cmd.sh b/examples/whisper/cmd.sh
@@ -4,13 +4,13 @@
 
 set -eou pipefail
 
-# take stdin and buffer it into a temp file
-input_temp=$(mktemp /tmp/whisper-input-XXXXXX)
-cat > "$input_temp"
+BASE_URL=$(echo "$1" | xargs dirname)
 
-# make sure we have a 16kHz WAV file
 output_file="${input_temp}_16khz.wav"
-ffmpeg -hide_banner -loglevel error -i "$input_temp" -ar 16000 -ac 1 "$output_file" 
+
+# replace relative *.ts URLs with the absolute URL to them
+cat | sed 's|^\([^#].*\)|'"$BASE_URL"'/\1|' \
+  | ffmpeg -protocol_whitelist https,fd,tls,tcp -f hls -i - -vn -acodec pcm_s16le -ar 16000 -ac 2 "$output_file"
 
 # generate the VTT file
 /app/main \

diff --git a/examples/whisper/scyllaridae.yml b/examples/whisper/scyllaridae.yml
@@ -4,3 +4,5 @@ allowedMimeTypes:
 cmdByMimeType:
   default:
     cmd: /app/cmd.sh
+    args:
+      - "%source-mime-uri"