# Mandarin pronunciation audio credits

Hanzi Arcade vendors the exact syllable recordings it uses from [hugolpz/audio-cmn](https://github.com/hugolpz/audio-cmn), specifically the `18k-abr/syllabs/` set. The local files are not hotlinked; `audio-manifest.js` records the selected filenames and `audio-cmn/README.md` is the upstream README retained beside them.

The upstream README describes the syllable set as 1,707 recordings by Chen Wang, licensed `CC-by-sa`, and credits Hugo Lopez, PLIDAM, and INALCO for project management, compression, and file renaming; Nicolas Vion is credited for recording software and technical support. The upstream README does not state a version number, so this project does not infer one. See the [upstream license section](https://github.com/hugolpz/audio-cmn#license), the [upstream README](https://github.com/hugolpz/audio-cmn/blob/master/README.md), or the [retained local README](audio-cmn/README.md) for the authoritative terms and credits.

The associated HSK word recordings in audio-cmn are credited separately to Yue Tan under `CC-by-sa` and are not bundled here. Their source notice is [packs.shtooka.net/cmn-caen-tan/readme.txt](http://packs.shtooka.net/cmn-caen-tan/readme.txt).

Hanzi Arcade’s native player uses the local syllable files as a deterministic pinyin-to-asset lookup. If a syllable is not present in the upstream set (notably some neutral-tone spellings), playback falls back to the selected Computer Voice so the app never substitutes a known wrong tone.