This project convert a folder or a RSS URL to Studio pack zip for Lunii device, see file structure below.
Supported OS: Windows / Linux / macOS
⭐ Une grosse communauté est présente sur Discord pour créer et partager des pack Lunii ! ⭐
⭐ A big french community is present on Discord to create and share Lunii packs ⭐
🔊💬🎵 Other devices/Apps use the Studio pack format :
- Telmi-story-teller for the Miyoo Mini retro gaming console
- Conty - android app on the PlayStore
- Boîte à histoires - android app apk
- Nimilou - android app on the PlayStore
- Grigri - the open storyteller
- open-story-teller - Open source hardware/software
studio-pack-generator "my story folder OR a RSS URL"will generate "my story folder-xxxxxxxxxx.zip" that can be imported in Studio
Examples:
studio-pack-generator http://radiofrance-podcast.net/podcast09/rss_19721.xmlstudio-pack-generator "Musics"studio-pack-generator "Encore une histoire"
Windows release of studio-pack-generator embeds these tools in zip file, and use Windows TTS instead of picoTTS (unless you have WSL and picoTTS installed).
- ffmpeg : used to extract images from story mp3 files, increase volume of
files, convert to the right format.
→ Use--skip-audio-convertand--skip-extract-image-from-mp3to avoid this usage. - imagemagick : used to generate menu image files.
→ Use--skip-image-item-gento avoid this usage. - picoTTS : used to generate menu audio files.
→ Use--skip-audio-item-gento avoid this usage.
Install optional dependencies :
sudo apt update && sudo apt install -y ffmpeg libttspico-utils imagemagick
Use "-miva" option to skip all generations that use these tools.
Install binary from release page and run it :
studio-pack-generator-x86_64-linux "my story folder or a rss url"
or studio-pack-generator-x86_64-windows.exe "my story folder or a rss url"
or studio-pack-generator-aarch64-apple "my story folder or a rss url"
or studio-pack-generator-x86_64-apple "my story folder or a rss url"
Or clone the repo and run with Deno :
This project is written in Typescript for deno runtime. Install deno : https://deno.land/
git clone https://github.com/jersou/studio-pack-generator
cd studio-pack-generator
deno -A studio_pack_generator.ts "my story folder or a rss url"
deno -A jsr:@jersou/studio-pack-generator "my story folder or a rss url"
Simplest example, only 1 menu level, without audio/image of menus/items :
📂 Story folder
└── 📂 Choose a story ← 📂 first menu
├── 🎵 the story 1.mp3 ← 📗 audio story
├── 🎵 the story 2.mp3 ← 📗 audio story
└── 🎵 the story 3.mp3 ← 📗 audio storySimple example, 2 levels of menus, without audio/image of menus/items :
📂 Story folder
└── 📂 Choose a character ← 📂 first menu
├── 📂 Alice ← 📂 first choice of the first menu
│ └── 📂 Choose a place ← 📂 second menu
│ ├── 🎵 the city.mp3 ← 📗 audio story
│ └── 🎵 the jungle.mp3 ← 📗 audio story
└── 📂 Bob ← 📂 second choice of the first menu
└── 📂 Choose a place ← 📂 second menu
├── 🎵 the desert.mp3 ← 📗 audio story
└── 🎵 the jungle.mp3 ← 📗 audio storystudio-pack-generator will generate menu files, they could be manually overwritten, and the next studio-pack-generator run will not regenerate these files :
📂 Story folder
├── 🎵 0-item.mp3 ← ⏩ story audio title, generated if missing
├── 🔳 0-item.png ← ⏩ story image title, generated if missing
├── 🔳 0-night-mode.mp3 ← ⏩ story audio night mode transition, generated if missing and if the mode is enable
└── 📂 Choose a character ← 📂 first menu
├── 🎵 0-item.mp3 ← ⏩ audio menu, generated if missing
├── 📂 Alice ← 📂 first choice of the first menu
│ ├── 🎵 0-item.mp3 ← ⏩ audio choice, generated if missing
│ ├── 🔳 0-item.png ← ⏩ image choice, generated if missing
│ └── 📂 Choose a place ← 📂 second menu
│ ├── 🎵 0-item.mp3 ← ⏩ audio menu, generated if missing
│ ├── 🔳 0-item.png ← ⏩ audio menu, generated if missing
│ ├── 🎵 the city.item.mp3 ← ⏩ audio story title, generated if missing
│ ├── 🔳 the city.item.png ← ⏩ image story title, generated if missing
│ ├── 🎵 the city.mp3 ← 📗 audio story
│ ├── 🎵 the jungle.item.mp3 ← ⏩ audio story title, generated if missing
│ ├── 🔳 the jungle.item.png ← ⏩ image story title, generated if missing
│ └── 🎵 the jungle.mp3 ← 📗 audio story
└── 📂 Bob ← 📂 second choice of the first menu
├── 🎵 0-item.mp3 ← ⏩ audio choice, generated if missing
├── 🔳 0-item.png ← ⏩ image choice, generated if missing
└── 📂 Choose a place ← 📂 second menu
├── 🔳 0-item.mp3 ← ⏩ audio menu, generated if missing
├── 🔳 0-item.png ← ⏩ audio menu, generated if missing
├── 🎵 the desert.item.mp3 ← ⏩ audio story title, generated if missing
├── 🔳 the desert.item.png ← ⏩ image story title, generated if missing
├── 🎵 the desert.mp3 ← 📗 audio story
├── 🎵 the jungle.item.mp3 ← ⏩ audio story title, generated if missing
├── 🔳 the jungle.item.png ← ⏩ image story title, generated if missing
└── 🎵 the jungle.mp3 ← 📗 audio storyThere is no limit to the nesting of menus, for example :
📂 Story folder
└── 📂 Choose a character ← 📂 first menu
├── 📂 Alice ← 📂 first choice of the first menu
│ └── 📂 Choose a place ← 📂 second menu
│ └── 📂 Building ← 📂 second choice of the first menu
│ │ └── 📂 Choose the floor ← 📂 third menu
│ │ ├── 🎵 the floor 1.mp3 ← 📗 audio story
│ │ └── 🎵 the floor 2.mp3 ← 📗 audio story
│ ├── 🎵 the city.mp3 ← 📗 audio story : mix menus/stories is possible
│ └── 🎵 the jungle.mp3 ← 📗 audio story : mix menus/stories is possible
├── 🎵 Bob.mp3 ← 📗 audio story : mix menus/stories is possible
...Since v0.1.11.
studio-pack-generator can embed zip studio packs in the tree structure :
📂 Story folder
└── 📂 Choose a character ← 📂 first menu
├── 📦 Alice.zip ← 📦 pack as menu entry
├── 🎵 Bob.mp3 ← 📗 audio story
...The "super pack" will look like :
📂 Story folder
└── 📂 Choose a character ← 📂 first menu
├── 📂 Alice ← 📂 The Alice.zip pack
│ └── 📂 Choose a place ← 📂 second menu
│ └── 📂 Building ← 📂 second choice of the first menu
│ │ └── 📂 Choose the floor ← 📂 third menu
│ │ ├── 🎵 the floor 1.mp3 ← 📗 audio story
│ │ └── 🎵 the floor 2.mp3 ← 📗 audio story
│ ├── 🎵 the city.mp3 ← 📗 audio story
│ └── 🎵 the jungle.mp3 ← 📗 audio story
├── 🎵 Bob.mp3 ← 📗 audio story
...- The first digit of file/folder name are ignored, it's useful to sort stories/menus.
- To keep numbers in generated items : "- 3 petits cochons.mp3" or "12 - 3 petits cochons.mp3".
- Image formats : png, jpg, bmp.
- Audio formats : mp3, ogg, opus, wav.
To run the GUI, use --gui : studio-pack-generator --gui story-path-here
The GUI does not work in RSS mode. This mode serve a web app on http://localhost:5555/
Usage: studio-pack-generator [options] [--] <story path | RSS URL> convert a folder or RSS url to Studio pack
Options:
-h, --help Show this help [default: false]
--config The json config file [string]
-d, --add-delay add 1 second at the beginning and the end of audio files [default: false]
-n, --auto-next-story-transition go to next story of group at end of stories [default: false]
-b, --select-next-story-at-end select the next story in the menu at end [default: false]
-l, --lang the lang used to generate menu and items. Auto detected by default [default: ""]
-t, --night-mode enable night mode : add transitions to an uniq endpoint [default: false]
-o, --output-folder zip output folder [string]
-c, --seek-story cut the beginning of stories: 'HH:mm:ss' format or 'N' sec [string]
-v, --skip-audio-convert skip convert audio (and skip increase volume) [default: false]
-j, --skip-image-convert skip image convert [default: false]
-a, --skip-audio-item-gen skip audio item generation [default: false]
-m, --skip-extract-image-from-mp-3 skip extract item image from story mp3 [default: false]
-i, --skip-image-item-gen skip image item generation [default: false]
--image-item-gen-font font used for image item generation [default: "Arial"]
--thumbnail-from-first-item gen thumbnail from first item instead of first chapter [default: false]
-s, --skip-not-rss skip all except download RSS files [default: false]
--rss-split-length RSS will be split in parts of N length [default: 10]
--rss-split-seasons RSS create different packs per season [default: false]
--rss-episode-numbers add RSS episode number to stages [default: false]
--rss-min-duration RSS min episode duration [default: 0]
--rss-use-subtitle-as-title Use rss items subtitle as title [default: false]
--rss-use-image-as-thumbnail Use rss image (first item with image) as thumbnail [default: false]
--use-thumbnail-as-root-image Use thumbnail as 'root' image instead of generated one [default: false]
-r, --skip-rss-image-dl skip RSS image download of items [default: false]
-w, --skip-wsl disable WSL usage [default: false]
-z, --skip-zip-generation only process item generation, don't create zip [default: false]
-e, --use-open-ai-tts generate missing audio item with Open AI TTS [default: false]
-k, --open-ai-api-key the OpenAI API key [string]
-g, --open-ai-model OpenAi model : tts-1, tts-1-hd [default: "tts-1"]
-p, --open-ai-voice OpenAi voice : alloy, echo, fable, onyx, nova, shimmer [default: "onyx"]
--use-coqui-tts use coqui TTS [default: false]
--coqui-tts-use-cuda enable CUDA in coqui TTS [default: false]
--coqui-tts-model coqui TTS model [default: "tts_models/multilingual/multi-dataset/xtts_v2"]
--coqui-tts-language-idx coqui TTS language_idx [default: "fr"]
--coqui-tts-speaker-idx coqui TTS speaker_idx [default: "Abrahan Mack"]
-x, --extract extract a zip pack (reverse mode) [default: false]
--extract-disable-night-mode disable night mode in extract mode [default: false]
-u, --gui open GUI (on localhost:5555) [default: false]
--port port of GUI server [default: 5555]
--skip-read-tts-cache disable the TTS cache usage [default: false]
--skip-write-tts-cache disable the TTS cache write [default: false]
--tts-cache-path path to the TTS cache [default: "<Studio-Pack-Generator dir>/.spg-TTS-cache"]
--custom-script custom script to be used for custom image... handling [string]
--metadata Metadata of the pack [object]
--i-18-n Custom i18n [object]
Separate options by spaces, ex :
- short version :
studio_pack_generator -v -j -a "the story"orstudio_pack_generator -vja "the story" - long version :
studio_pack_generator --skip-audio-convert --skip-image-convert --skip-audio-item-gen "the story"
- Generate studio pack from file tree.
- Generate menu image/audio file if missing.
- Extract image from mp3 file as story image in menu.
- Increase audio volume of stories if needed.
- Download podcast from a RSS url and generate the story tree, cut by parts of 10 stories.
- Convert mp3 files to right format (mp3, 44100 Hz, mono).
- Convert image files to right format (320x240).
- Generate story thumbnail.
- Option to chaining the stories.
- Option enable the night mode.
- Option to add 1 sec of silence at the beginning and end of sound files.
- Option to skip the beginning of stories.
- Zip Pack aggregation
- OpenAI & Coqui TTS
- a GUI
If the file metadata.json exists in the story folder, it will be used to
overwrite the story.json metadata.
All key/value are optional, ex:
{
"title": "title - overwrite",
"description": "description - overwrite",
"format": "v1",
"version": 1,
"nightMode": false
}A folder <studio-pack-generator install dir>/.spg-TTS-cache/ is used to keep
the generated audio files.
To use OpenAI TTS, use --use-open-ai-tts option, and you must set the API key:
- set OPENAI_API_KEY in the environnement variables
- or use --open-ai-api-key parameter
- or enter the key when the program prompt
Coqui is "a deep learning toolkit for Text-to-Speech": https://github.com/idiap/coqui-ai-TTS
To install coqui
: pip install coqui-tts
Extract a file stucture from zip pack :
-x, --extract extract a zip pack (reverse mode) [boolean] [default: false]Example :
studio-pack-generator -x 2-full.zipor
studio-pack-generator -x -o output/dir 2-full.zipNote: it doesn't work well with "menu" nodes and with pack without "question" stage.
To speed up / save CPU
Usage : --custom-script=<path>
export interface CustomModule {
fetchRssItemImage?: (item: RssItem, opt: StudioPackGenerator) => Promise<string>;
fetchRssItemTitle?: (item: RssItem, opt: StudioPackGenerator) => Promise<string>;
}
The parameters can be imported from a json file with :
--config-file=<json file path>
File format (all the properties are optionals) :
{
"addDelay": false,
"autoNextStoryTransition": false,
"selectNextStoryAtEnd": false,
"nightMode": false,
"skipAudioConvert": false,
"skipImageConvert": false,
"skipAudioItemGen": false,
"skipExtractImageFromMp3": false,
"skipImageItemGen": false,
"skipNotRss": false,
"skipRssImageDl": false,
"skipWsl": false,
"skipZipGeneration": false,
"useOpenAiTts": false,
"lang": "fr-FR",
"outputFolder": "/tmp/",
"seekStory": "1",
"openAiApiKey": "",
"openAiModel": "tts-1",
"openAiVoice": "onyx",
"gui": false,
"imageItemGenFont": "Arial",
"thumbnailFromFirstItem": false,
"rssSplitLength": 10,
"rssSplitSeasons": false,
"rssMinDuration": 0,
"rssUseImageAsThumbnail": false,
"useThumbnailAsRootImage": false,
"rssEpisodeNumbers": false,
"useCoquiTts": false,
"coquiTtsModel": "tts_models/multilingual/multi-dataset/xtts_v2",
"coquiUseCuda": false,
"coquiTtsLanguageIdx": "fr",
"coquiTtsSpeakerIdx": "Abrahan Mack",
"port": 5555,
"skipWriteTtsCache": false,
"skipReadTtsCache": false,
"ttsCachePath": "/tmp/spg-tts-cache",
"i18n": {
"special": "Special",
"season": "Season %d",
"storyQuestion": "Choose your story",
"partQuestion": "Choose your part",
"NightModeTransition": "Want to listen to a new story?"
}
}At the end of the generation, a file 0-config.json will be written in the
story folder. It can be use for the next run :
studio_pack_generator --config-file="<story path>/0-config.json" "<story path>"
Some dev command are listed in the deno.json file :
- fmt: format the code
- gen-bin: generate the binaries
- gen-cov: generate the test coverage
- check: deno check studio_pack_generator.ts
- lint: lint the code
- pre-commit: fmt && lint && test && check
- start: run studio_pack_generator.ts
- test: launch tests
- test-watch: launch tests on file change
Usage : deno task <command>, ex : deno task fmt
Note: some dependencies are vendored in vendor/ folder, it's to publish on JSR
(which does not allow http imports).
- #19 : end node
- clean file download option unless -z
- use https://github.com/rhasspy/piper for TTS
- use git LFS
- ...