Content Pack Format Reference

Worth knowing

• Since 0.5.0 there is an in-game metadata editor: main menu → Extras → "Edit Voice Packs & Clip Metadata". There you can assign images per sample, add captions, tag samples, credit contributing authors (players can search packs by author) and attach a readme.
• A pack is named after its folder by default; the editor lets you change the displayed name.
• Each clip can carry a same-named .txt sidecar using caption= for the line prompt and image= for a specific image.

Worth knowing

• Why OGV is mandatory: the engine the game is built on currently supports only that one open video format — MP4 and AVI are not options. That is the whole reason for the requirement.
• Timestamps are entered by hand. The developer’s workflow is to set Audacity’s audio position readout to "hh:mm:ss + milliseconds" and write the seconds into the export file name (e.g. 07_MyClip_44-048), then fill them into the editor afterwards.
• Dub Mode rules: you get unlimited retakes, but you are not allowed to hear your own recordings until the game ends. Once everything has been performed, the takes are played back dubbed over the video.
• Scenes with several audio sources can be tagged with "dub character" groups so players can filter for the characters they want to perform. Unselected characters use their original audio on playback.
• Clips can be toggled "dub only", which keeps them out of every mode except Dub Mode.

Worth knowing

• The image is positioned so its bottom edge touches the studio floor. To shift the character upward, add empty space at the bottom.
• Nine events can carry audio: being introduced by the host, winning, losing, and getting a judge score of 0 / 1 / 2 / 3 / 4 / 5. That list is also the clearest confirmation that scoring runs 0–5 rather than on a percentage scale.
• Name, the two theme colours and the host introduction can all be edited in the in-game Customization menu.

Worth knowing

• Judge names are edited in the Customization menu.
• To silence the score blips — for example if you only want judge voices — add play_voices_with_blips set to false directly in the config JSON.

Worth knowing

• Creating a studio folder in-game generates a template glb showing the size and position of the judges, contestants and three screens. Using it to align your model is the easy path — nothing in the template is required for the model to work.
• The bigger the model, the longer the load. Studios with a lot of vertices can take several seconds to load into the game.
• Music loops the whole track by default. music_studio_loop_start sets a loop point: for WAV this is the starting sample, for MP3 and OGG the starting time in seconds.
• The studio includes its own lighting. If your model brings its own, set use_builtin_light to false in the studio config.
• The colours of the audio recording overlay are edited in the in-game Customization menu.

Worth knowing

• Dialogue is grouped by stage. The shipped structure is match_singleplayer.intro.{a_welcome, b_contestant, c_judges, d_explanation} and match_singleplayer.round.{b_post_record, c_post_listen, round_next}, and every entry is an array of strings.
• Lines support placeholders: <host_name> and <player> are replaced with the actual host and contestant names.
• The stage names alone describe the round flow: recording finishes (b_post_record) → the take is listened back (c_post_listen) → the judges vote → the next round (round_next).
• Because each stage is an array, you can write several lines and let the game rotate between them.

Worth knowing

• config_menu.json groups into three blocks: audio, background and ui.
• background holds image.{use_type, scroll.x, scroll.y}, overlay.on, clip_disc.{state, color}, top_gradient, bottom_gradient, circles, waves and letterbox (with both color and accent).
• ui.button has invert / color1 / color2. All colours are hex strings, some carrying alpha — for example abe5f5cc.
• music_menu_loop_start follows the same rule as the studio: WAV uses the starting sample, MP3 and OGG use seconds.

Worth knowing

• broad_keywords fire when the keyword appears anywhere in the viewer’s first word. So clap also catches Clap / clap2 / clapping / CLAPS / cccclappp.
• exact_keywords only fire on an exact match, and they are case-sensitive — which makes them the right choice for emote names.
• The syntax matches other content packs: a bracketed section name, then the audio file on the left and an array of keywords on the right, e.g. clap.ogg = ["clap", "👏"].