Credits and licenses

This tool was built by Kushal Sapkota at AAFAI AI. It stands on the work of many open projects and generous people. Thank you to everyone listed here.

Voices

Used for	Voice	Source and license
English boy and monkey	Ryan	RyanSpeech dataset by Rohola Zandie and others CC BY-NC-SA 4.0Non-commercial license. Source
English girl	Amy	Mimic 3 voices by Mycroft AI See sourcegithub.com/MycroftAI/mimic3-voices
English robot	Joe	OHF-Voice voice datasets CC0Source
Nepali, Voice 1	Chitwan	OHF-Voice voice datasets CC0Source
Nepali, Voice 2	Google	Crowdsourced high-quality Nepali speech data set, created by Google (OpenSLR 43) CC BY-SA 4.0Credit required. openslr.org/43

The voices run on Piper, an open text-to-speech engine from the Open Home Foundation. Piper's voices were fine-tuned from a base voice trained on the Lessac Blizzard 2013 data from the University of Edinburgh. Please read the voice licenses above before using a video commercially.

Software

Tool	What it does here	License
Piper (piper-tts)	Speaks the script	GPL-3.0
ONNX Runtime	Runs the voices	MIT
FFmpeg	Builds the video and mixes the sound	LGPL / GPL, ffmpeg.org
Cairo and pycairo	Draws the characters and scenes	LGPL / MPL
Pillow	Handles pictures and Nepali text shaping	MIT-CMU
FastAPI and Uvicorn	Runs this website	MIT and BSD-3

Fonts

Bricolage Grotesque and DM Sans (website text), via Google Fonts, SIL Open Font License 1.1
Noto Sans Devanagari (Nepali text inside videos), by Google, SIL Open Font License 1.1
DejaVu Sans (English text inside videos), free license based on Bitstream Vera

Made for this site

The characters, backgrounds, background music and sound effects are drawn and generated by code written for this site. They are not copied from other works.

Corrections

If you created something listed here and want the credit changed, or you spot a mistake, please write to kushalsapkota@aafaiai.com.