- Joined
- Nov 7, 2021
That clip channel kind of sucks, lazy ~10 min clips of random shit.
The telegram clip channel https://web.telegram.org/k/#@topnarezki seems to have better short clips. Here are some random ones with subtitles
i (chatgpt) made a shell script to transcribe youtube videos/mp4s with OpenAI's Whisper and burn the subtitles on with ffmpeg. its below in case someone's interested
The telegram clip channel https://web.telegram.org/k/#@topnarezki seems to have better short clips. Here are some random ones with subtitles
i (chatgpt) made a shell script to transcribe youtube videos/mp4s with OpenAI's Whisper and burn the subtitles on with ffmpeg. its below in case someone's interested
bash
python3
pip
ffmpeg
git
script's --setup will install the python dependencies:
torch, torchvision, torchaudio, openai-whisper, yt-dlp, ffmpeg-python
python3
pip
ffmpeg
git
script's --setup will install the python dependencies:
torch, torchvision, torchaudio, openai-whisper, yt-dlp, ffmpeg-python
Bash:
#!/bin/bash
set -e
LANGUAGE="ru"
INPUT=""
USE_LANGUAGE=true
MODEL="medium"
# Track timing
declare -A TIMERS
to_minutes_seconds() {
local duration=$1
local minutes=$((duration / 60))
local seconds=$((duration % 60))
printf "%2dm %02ds" "$minutes" "$seconds"
}
start_timer() {
TIMERS[$1]=$(date +%s)
}
stop_timer() {
local name=$1
local end=$(date +%s)
local start=${TIMERS[$name]}
local duration=$((end - start))
TIMERS[$name]=$duration
printf "⏱️ %-15s: %s\n" "$name" "$(to_minutes_seconds $duration)"
}
check_tool() {
command -v "$1" >/dev/null 2>&1 || {
echo "❌ Missing: $1"
echo "👉 Please install it manually or run './translate_and_burn.sh --setup'"
exit 1
}
}
setup_dependencies() {
echo "🔧 Setting up Python dependencies..."
check_tool python3
check_tool curl
check_tool ffmpeg
python3 -m venv venv
source venv/bin/activate
pip install --upgrade pip
pip uninstall -y torch torchvision torchaudio || true
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip install git+https://github.com/openai/whisper.git
pip install yt-dlp ffmpeg-python
echo "✅ Dependencies installed."
exit 0
}
download_video() {
local url="$1"
echo "📥 Downloading YouTube video..."
yt-dlp -f bestvideo+bestaudio --merge-output-format mp4 "$url" -o "$VIDEO_OUTPUT"
if [[ ! -f "$VIDEO_OUTPUT" ]]; then
echo "❌ Download failed. File not found: $VIDEO_OUTPUT"
ls -la
exit 1
fi
echo "✅ Saved as: $VIDEO_OUTPUT"
}
run_whisper() {
echo "🧠 Transcribing and translating using Whisper..."
start_timer whisper
if [ "$USE_LANGUAGE" = true ]; then
whisper --task translate --model "$MODEL" --language "$LANGUAGE" -- "$VIDEO_OUTPUT"
else
whisper --task translate --model "$MODEL" -- "$VIDEO_OUTPUT"
fi
# Find any generated .srt file matching base name
SUB_FILE=$(ls "${VIDEO_OUTPUT%.*}"*.srt 2>/dev/null | head -n 1)
if [ -f "$SUB_FILE" ]; then
mv "$SUB_FILE" "$SUBTITLE_OUTPUT"
echo "✅ Subtitles saved as $SUBTITLE_OUTPUT"
else
echo "❌ Subtitle file not found. Whisper may have failed."
exit 1
fi
stop_timer whisper
}
burn_subtitles() {
echo "🔥 Burning subtitles into video using CUDA (h264_nvenc)..."
start_timer burn
ffmpeg -y -hwaccel cuda -i "$VIDEO_OUTPUT" -vf "subtitles=$SUBTITLE_OUTPUT" -c:v h264_nvenc -c:a copy "translated_${VIDEO_OUTPUT}"
echo "✅ Output: translated_${VIDEO_OUTPUT}"
stop_timer burn
}
# === Entry Point ===
if [[ "$1" == "--setup" ]]; then
setup_dependencies
fi
# Parse arguments
for arg in "$@"; do
case $arg in
--lang=*)
LANGUAGE="${arg#*=}"
USE_LANGUAGE=true
;;
http*)
INPUT="$arg"
;;
*.mp4)
INPUT="$arg"
;;
esac
done
if [[ -z "$INPUT" ]]; then
echo "❌ Usage: $0 <YouTube URL or .mp4 file>"
exit 1
fi
YOUTUBE_ID=$(echo "$INPUT" | grep -oE 'v=([a-zA-Z0-9_-]+)' | cut -d= -f2)
BASE_NAME="${YOUTUBE_ID:-$(basename "$INPUT" .mp4)}"
VIDEO_OUTPUT="${BASE_NAME}.mp4"
SUBTITLE_OUTPUT="${BASE_NAME}.srt"
check_tool yt-dlp
check_tool whisper
check_tool ffmpeg
# Download or copy
if [[ "$INPUT" == http* ]]; then
download_video "$INPUT"
else
if [[ "$INPUT" != "$VIDEO_OUTPUT" ]]; then
cp "$INPUT" "$VIDEO_OUTPUT"
echo "✅ Copied local file to $VIDEO_OUTPUT"
else
echo "ℹ️ Input file already matches expected name: $VIDEO_OUTPUT"
fi
fi
start_timer total
run_whisper
burn_subtitles
stop_timer total
echo "============================="
echo "🎬 Done. Summary of durations:"
stop_timer whisper
stop_timer burn
stop_timer total
"now that you've seen some random Russian agree with me - maybe you want to change your opinion???"
