VideTShrts
← All workflowsOpen source · runs locally

A faceless Short from a script, with open-source tools

Write a script, voice it with Piper text-to-speech, lay it over your own footage or generated B-roll, and add captions with Whisper. Everything runs on your computer.

Cost
Free
Time
About 30 minutes per Short
Level
Intermediate (command line)
Hardware
Any recent computer for voice and captions; a strong GPU only if you generate B-roll locally.

What you need

  • Piper: Local text-to-speech voices
  • Whisper: Captions timed to the voice
  • FFmpeg: Combine voice, footage and captions
  • Wan 2.2: Alibaba model, open weights
  • LTX-2.5: Lightricks model, open weights

Best for: Explainers and listicles, Channels that don't show a face, Testing ideas quickly.

Steps

  1. 1

    Write a short script

    Aim for 120 to 150 words for about a minute. Put the hook in the first sentence and keep one idea per sentence. Save it as script.txt.

  2. 2

    Install Piper and download a voice

    Listen to the voice samples first; each voice has its own license on its model card.

    pip install piper-tts
    python3 -m piper.download_voices en_US-lessac-medium
  3. 3

    Generate the voiceover

    --input-file reads the script; --sentence-silence adds a short pause between sentences.

    python3 -m piper -m en_US-lessac-medium --input-file script.txt --sentence-silence 0.3 -f voice.wav
  4. 4

    Get background footage

    Use your own clips or stock you have rights to, or generate 5-second B-roll with an open model (see our Wan 2.2 in ComfyUI workflow). A vertical 1080x1920 clip works best.

  5. 5

    Combine footage and voice

    -stream_loop -1 repeats the background until the voice ends; -shortest stops at the end of the voice.

    ffmpeg -stream_loop -1 -i broll.mp4 -i voice.wav -map 0:v -map 1:a -vf "scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920" -c:v libx264 -pix_fmt yuv420p -c:a aac -shortest draft.mp4
  6. 6

    Caption it

    Transcribe the draft so captions match the voice exactly, then burn them in.

    whisper draft.mp4 --model small --language en --word_timestamps True --max_line_width 24 --max_line_count 1 --output_format srt
    ffmpeg -i draft.mp4 -vf "subtitles=draft.srt:force_style='FontName=Arial,FontSize=16,Bold=1,Outline=3,Alignment=2,MarginV=110'" -c:a copy final.mp4

Questions

Can I use the Piper voices commercially?

It depends on the voice. Piper itself is open source, and each voice model lists its own license and training data on its model card. Check it before monetizing.

Does YouTube allow AI voiceovers?

Platforms have their own rules on synthetic media and disclosure. Check the current policy of each platform you post to.

Sources

Commands checked against each tool's docs on November 19, 2026.

Get new workflows by email

One short email a week: new video models and tools, free-plan and price changes, and what we found worth using.

No spam. Reply "unsubscribe" to any email and you're off the list.