Credits & licenses

Little Sprout brings together original playroom artwork and music, open-source fonts and icons, and prerecorded synthetic teaching voices. These are the people and projects behind those materials.

Artwork & music

The game illustrations, cartoon teachers, classroom environments and handwriting graphics are original vector artwork created for Little Sprout. The app icon is an export of its original vector design.

The music and sound effects are original synthesized compositions. Animal sounds are playful electronic sound associations; they do not use sampled animal recordings, commercial songs or licensed loops. Emoji use the fonts supplied by your device.

Fonts & icons

Synthetic teaching voices

Spoken lessons are prerecorded, AI-generated speech in English, Hindi, Arabic, Spanish, Mandarin Chinese and Japanese. Each language has two recorded base voices. Cartoon teachers apply different pitch and pace settings to those recordings. The app does not record a child, clone a custom human reference voice or contact a speech service during play.

Kokoro

English, Hindi, Spanish, Mandarin Chinese and Japanese recordings use Hexgrad's Kokoro-82M v1.0. Its source, model and preset files are distributed under the Apache License 2.0. See the model credits, voice inventory, Kokoro source and Misaki pronunciation project.

Presets used for Little Sprout's recordings
LanguageFemale presetMale preset
Englishaf_heartam_michael
Hindihf_alphahm_omega
Spanishef_doraem_alex
Mandarin Chinesezf_xiaoxiaozm_yunjian
Japanesejf_alphajm_kumo

Japanese preset source credit

Kokoro's jm_kumo entry credits Koniwa / Television Nishinippon (TNC) for 「蜘蛛の糸」 (Kumonoito / The Spider's Thread), a reading of Ryūnosuke Akutagawa's public-domain text. Koniwa identifies the TNC announcer recordings as Creative Commons Attribution 3.0. See the Koniwa collection and credits, credited work and TNC reading project.

Little Sprout uses the resulting synthetic preset to speak newly authored teaching text and modifies pitch and pace for its characters. It does not redistribute the original reading. No endorsement by these contributors is implied. Kokoro also credits SIWIS (CC BY 4.0) in its overall training data; the SIWIS voice pack is not selected here.

Arabic voices

Arabic recordings use SILMA TTS v1, by SILMA AI. Its weights use the Apache License 2.0, and its inference code uses the MIT license. SILMA uses the F5-TTS architecture with its own pretrained weights. See the SILMA model card and SILMA source.

The vocoder is Charactr's Vocos mel-24kHz, distributed under the MIT license. Vocos model. Arabic conditioning uses a newly synthesized Kokoro am_michael reference, rather than a human recording. The Arabic female base is derived from the male rendition by changing pitch and formants while preserving its pace; it is not a separate female model preset.

App software

Little Sprout also uses open-source software libraries. Their copyright notices and license terms are available in the runtime software notices. The notices here do not imply endorsement of Little Sprout by any contributor or project.