TivalVideo Offline 0.3.0

Create narrated vertical Shorts or landscape videos from images, entirely offline after the first setup. TivalVideo uses a managed FFmpeg executable for video rendering and Piper for local AI narration.

Version 0.3.0 adds a downloadable narrator catalogue. Choose between American and British Piper voices, then create videos fully offline.

Install

python3 -m venv .venv
source .venv/bin/activate
pip install .

pip automatically installs FFmpeg and Piper dependencies. The first video automatically downloads the approximately 61 MB Piper voice model, or you can prepare it explicitly:

tivalvideo setup

After setup, disconnect from the internet and create videos offline.

Narrator voices

See all included voices and which ones are installed:

tivalvideo voices list

Install additional voices only when you need them:

tivalvideo voices install en_US-amy-medium
tivalvideo voices install en_US-joe-medium
tivalvideo voices install en_GB-alan-medium

Each medium voice is approximately 63 MB and stays cached for offline use.

Choose a voice in Python:

from tivalvideo import create_video

create_video(
    typing_text="The history of Python",
    narration="Python was created by Guido van Rossum.",
    voice="en_GB-alan-medium",
    background="midnight",
    output="python-history.mp4",
)

Available narrator keys:

• en_US-lessac-medium (default, American English)
• en_US-amy-medium (American English)
• en_US-joe-medium (American English)
• en_GB-alan-medium (British English)

The selected voice downloads automatically on first use. Remove an unused voice with:

tivalvideo voices remove en_US-amy-medium

Python example

from tivalvideo import create_video

create_video(
    images=["lesson1.png", "lesson2.png", "lesson3.png"],
    narration="Learn TivalTube with these simple examples.",
    output="tivaltube-short.mp4",
)

Auto-typing video

Create a complete Short without supplying lesson images:

from tivalvideo import create_video

create_video(
    typing_text="Welcome to my Python lesson!",
    narration="Welcome to my Python lesson.",
    background="ocean",
    output="typing-short.mp4",
)

Built-in background choices are midnight, ocean, sunset, and paper.

Use your own background image:

create_video(
    typing_text="Learn Python step by step",
    audio="my-recorded-voice.mp3",
    background_image="my-background.jpg",
    output="custom-short.mp4",
)

Recorded voice and music

Use recorded voice instead of AI narration and optionally mix in music:

create_video(
    images=["lesson1.png", "lesson2.png"],
    audio="voice.wav",
    music="background.mp3",
    output="lesson.mp4",
)

Do not pass both narration and audio; they are two alternative voice sources.

Terminal example

Create narration.txt, then run:

tivalvideo create \
  lesson1.png lesson2.png lesson3.png \
  --narration narration.txt \
  --output short.mp4

Use --landscape for a 1920×1080 video:

tivalvideo create *.png -n narration.txt -o youtube.mp4 --landscape

Features

• 1080×1920 TikTok and YouTube Shorts
• 1920×1080 landscape videos
• Offline Piper AI narration
• Managed FFmpeg binary—no apt install ffmpeg required
• Blurred full-screen background with centred images
• Automatic fade transitions
• H.264 video and AAC audio
• Automatic voice-model caching
• Selectable American and British narrator voices
• Voice list, install, and remove commands
• Auto-typing text animation
• MP3/WAV recorded voice and background music
• Built-in and user-provided auto-typing backgrounds
• Python API and terminal command

Notes

• The first setup requires internet access for the voice download.
• Generated videos work from local image files and text only.
• Keep the voice model cached to remain fully offline.

Powered by tivalsdeveloper.