Credits
Made by
- Daisy: creator, story, characters and direction. Music by Daisy Dark.
- Built with two AI assistants working under Daisy's direction: Cedar (the park's code, the server and the tests) and Oak (the 3D park, the characters and the renders). See How we use AI.
Voices
The characters speak with computer voices made on our own server by Piper (MIT License), from voice models trained on recordings that volunteers read for public-domain audiobooks.
- Ranger Kara: Kokoro af_heart (Kokoro's credit is below).
- Greg: Piper en_US-libritts_r-medium, speaker 35 (LibriTTS-R reader 4595).
- Bean and Lexie the pony: Piper en_US-libritts_r-medium, speaker 54 (LibriTTS-R reader 16); Bean's lines are raised in pitch.
- Sofia and Lucia, the Our World show hosts: Piper en_US-libritts_r-medium, speakers 281 and 1 (LibriTTS-R readers 2775 and 8699).
- HAL (HAL-9001 at the Pelican and on the Sky Ride; HAL-9002 and HAL-9003, the race hosts): one voice, Piper en_US-libritts_r-medium, speaker 805 (LibriTTS-R reader 17), with a light robot touch we add.
- The LibriTTS-R voices come from LibriTTS-R by Yuma Koizumi, Heiga Zen, Shigeki Karita, Yifan Ding, Kohei Yatabe, Nobuyuki Morioka, Michiel Bacchiani, Yu Zhang, Wei Han and Ankur Bapna (2023), licensed under CC BY 4.0. It's built on LibriTTS (Zen et al., 2019) and on LibriVox recordings, which are in the public domain. Thank you to the LibriVox readers.
- Bean's own voice is being chosen from our voice sheet; it will be named here when it's picked.
In 20 languages besides English, Ranger Kara, Greg and HAL speak with the voices below. In some languages only some of them have a recorded voice so far; their other lines, and any line without a recording, are read by your device's own voice.
- Kokoro by hexgrad (Apache License 2.0), run with kokoro-onnx (MIT License): Kara in English (af_heart), Spanish (ef_dora), Portuguese (pf_dora) and French (ff_siwis); Greg in Spanish (em_alex) and Portuguese (pm_alex); HAL in Spanish (em_santa) and Portuguese (pm_santa); Kara in Italian (if_sara) and Greg in Italian (im_nicola). The French voice comes from the SIWIS French Speech Synthesis Database (Honnet, Lazaridis, Garner and Yamagishi, 2017), licensed under CC BY 4.0.
- Greg in French: Piper fr_FR-upmc-medium, speaker 1 ("pierre"), from the UPMC French recordings, licensed under CC BY-SA 4.0.
- Greg in German: Piper de_DE-thorsten-medium, from the Thorsten-Voice recordings by Thorsten Müller (CC0).
- Kara in German: Piper de_DE-mls-medium, speaker 153 (Multilingual LibriSpeech reader 4576); HAL in German: the same model, speaker 182 (reader 5406); HAL in French: Piper fr_FR-mls-medium, speaker 28 (reader 5525); Kara in Dutch: Piper nl_NL-mls-medium, speaker 10 (reader 12500); HAL in Dutch: the same model, speaker 41 (reader 11157). All of these are from Multilingual LibriSpeech by Vineel Pratap, Qiantong Xu, Anuroop Sriram, Gabriel Synnaeve and Ronan Collobert (2020), licensed under CC BY 4.0 and built on LibriVox recordings.
- Catalan: Kara is Piper ca_ES-upc_ona-medium, Greg and HAL ca_ES-upc_pau-x_low, from the FestCat speech corpora of the Universitat Politècnica de Catalunya (UPC), licensed under CC BY-SA 3.0 ES.
- Slovenian: Greg and HAL are Piper sl_SI-artur-medium, trained by ppisljar from the studio recordings in ARTUR 1.0 (Darinka Verdonik et al., 2023, CLARIN.SI, hdl 11356/1776); the voice is licensed under CC BY 4.0 and its dataset under CC BY 3.0.
- Vietnamese: Kara is Piper vi_VN-vais1000-medium, from the VAIS-1000 Vietnamese speech synthesis corpus, licensed under CC BY 4.0.
- HAL in Italian: Piper it_IT-riccardo-x_low, from the M-AILABS Speech Dataset, whose license reads: Copyright (c) 2017-2019 by the original creators @ M-AILABS with the following license: Redistribution and use in any form, including any commercial use, with or without modification are permitted provided that the following conditions are met: Redistributions of source data must retain the above copyright notice, this list of conditions and the following disclaimer. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this downloaded data, source-code or binary-code without specific prior written permission. THIS DATA IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE and/or DATA, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- Given to the public domain (CC0), with our thanks: Czech jirka, Hungarian anna and imre, Dutch pim, Polish gosia and darkman, Romanian mihai, Russian denis, Slovak lili and Ukrainian ukrainian_tts (speakers tetiana and mykyta), from the Open Home Foundation voice datasets; Danish talesyntese and Swedish nst (trained by KBLab at the National Library of Sweden), from recordings kept by Språkbanken at the National Library of Norway; Greek rapunzelina and Finnish harri, from single-speaker speech datasets shared on Kaggle; and Polish mc_speech, from the MC Speech dataset.
Software
- three.js r170, the 3D engine (MIT License), with its Draco, meshopt and Basis Universal decoders (Apache License 2.0 and MIT).
- three-mesh-bvh 0.8.3, fast collision and picking (MIT License).
- The grown-ups' pages translate with Firefox Translations models (Mozilla Public License 2.0), run on our own server.
- The 3D park is made in Blender.
- The signs use the Fredoka typeface by Milena Brandão and Hafontia (SIL Open Font License 1.1).
Standards we teach toward
- Head Start Early Learning Outcomes Framework (U.S. Office of Head Start, public domain).
- Common Core State Standards for Mathematics and English Language Arts, © 2010 National Governors Association Center for Best Practices and Council of Chief State School Officers.
- Next Generation Science Standards, NGSS Lead States (2013). NGSS is a registered trademark of WestEd; the standards' authors don't endorse Bean's Place.