StoryTeller

A multimodal AI story teller, built with Stable Diffusion, GPT, and neural text-to-speech (TTS).

Given a prompt as an opening line of a story, GPT writes the rest of the plot; Stable Diffusion draws an image for each sentence; a TTS model narrates each line, resulting in a fully animated video of a short story, replete with audio and visuals.

Installation

PyPI

Story Teller is available on PyPI.

$ pip install storyteller-core

Source

Clone the repository.

$ git clone https://github.com/jaketae/storyteller.git
$ cd storyteller

Install dependencies.

$ pip install .

Note: For Apple M1/2 users, mecab-python3 is not available. You need to install mecab before running pip install. You can do this with Hombrew via brew install mecab. For more information, refer to SamuraiT/mecab-python3#84.

(Optional) To develop locally, install dev dependencies and install pre-commit hooks. This will automatically trigger linting and code quality checks before each commit.

$ pip install -e .[dev]
$ pre-commit install

Quickstart

The quickest way to run a demo is through the CLI. Simply type

$ storyteller

The final video will be saved as /out/out.mp4, alongside other intermediate images, audio files, and subtitles.

To adjust the defaults with custom parametes, toggle the CLI flags as needed.

$ storyteller --help
usage: storyteller [-h] [--writer_prompt WRITER_PROMPT]
                   [--painter_prompt_prefix PAINTER_PROMPT_PREFIX] [--num_images NUM_IMAGES]
                   [--output_dir OUTPUT_DIR] [--seed SEED] [--max_new_tokens MAX_NEW_TOKENS]
                   [--writer WRITER] [--painter PAINTER] [--speaker SPEAKER]
                   [--writer_device WRITER_DEVICE] [--painter_device PAINTER_DEVICE]

optional arguments:
  -h, --help            show this help message and exit
  --writer_prompt WRITER_PROMPT
  --painter_prompt_prefix PAINTER_PROMPT_PREFIX
  --num_images NUM_IMAGES
  --output_dir OUTPUT_DIR
  --seed SEED
  --max_new_tokens MAX_NEW_TOKENS
  --writer WRITER
  --painter PAINTER
  --speaker SPEAKER
  --writer_device WRITER_DEVICE
  --painter_device PAINTER_DEVICE

Usage

For more advanced use cases, you can also directly interface with Story Teller in Python code.

Load the model with defaults.

from storyteller import StoryTeller

story_teller = StoryTeller.from_default()
story_teller.generate(...)

Alternatively, configure the model with custom settings.

from storyteller import StoryTeller, StoryTellerConfig

config = StoryTellerConfig(
    writer="gpt2-large",
    painter="CompVis/stable-diffusion-v1-4",
    max_new_tokens=100,
)

story_teller = StoryTeller(config)
story_teller.generate(...)

License

Released under the MIT License.

imomin / storyteller