Voices Sounds - Mimic Party Pack Guide

Voices is the third official themed pack bundled with Mimic Party: human vocal references — spoken lines, calls and other people-shaped sounds you must reproduce with the only instrument you have, your own voice. No official track list exists, so we keep to the theme. The draw works as always: each round's reference sound is picked at random across the packs you own, heard once by the whole table, then copied in one take with no second chance.

The official section on scoring is literally titled "We don't care about your voice": timbre is not scored, and neither is the microphone you bought. A deep voice and a high voice doing the same curve score the same. That design choice defines the whole pack — you are graded on Melody (the relative pitch curve), Rhythm and Attacks, never on how much you sound like the original speaker. Acting the impression harder does nothing; tracing the pitch movement does.

Because register never counts against you, imitating a voice far from your own is not the handicap it looks like. A bass-voiced player copying a squeaky line simply compresses the curve into a lower range — the relative pitch comparison keeps the score. The same logic means voice tricks and character voices are cosmetics: they change your timbre, which is ignored, and do nothing for the three dimensions that count. Win the curve, the spacing and the onset count instead.

Reference sounds are picked at random across the packs you own — so Voices pools with Animals, Machines and every Workshop pack your group installs. Everyone at the table needs the selected packs before the round starts; the lobby shows who is still downloading, and updated Workshop packs may need a quick re-subscribe to sync. The guides below cover building your own pack and exactly how the AI scores each take.

Because timbre and microphone quality are both ignored, performing the line louder or with more theatrical effort buys nothing — the AI hears pitch movement, timing and onset count, not commitment. What extra effort actually risks is pitch control: pushed voices go sharp, and a strained take wobbles through the very Melody curve you are being graded on. The players who consistently score well on voice references deliver at conversational volume, where their pitch control is at its most reliable. Save the acting for the table laugh — it is worth more there anyway.

Spoken-line references do not care whether you understand the words — the AI compares sound shapes, not language. That cuts both ways. A line in a language nobody at the table speaks is still fully scoreable: treat it as pure melody-plus-rhythm, hum the pitch contour with the syllable rhythm, and you can outscore a native speaker who tries to articulate the actual words. When a familiar line comes up, resist narrating it word-perfectly at the expense of the curve — the syllable onsets (Attacks) and the pitch shape (Melody) are where the points live, not the diction.