Guide · Strategy

How the Scoring System Works (and How to Score Higher)

Updated August 28, 2026 · ~5 min read

Most new players lose their first session and assume the judges are unfair. They are not. The scoring rewards specific, repeatable things, and once you internalize what tends to score higher, your round-by-round average climbs visibly within a single night. This guide covers what players have observed about the scoring system, why some impressions consistently outscore others, and the practical habits that pull your average up.

A note on scope: the developer has never published a formal spec for how the judging algorithm weighs each input, and the underlying logic is widely described in community discussion as opaque. What follows is what experienced players have noticed, not a leaked formula. If the developer ever publishes official scoring details, this guide should be re-read with that context.

What experienced players tend to see the judges rewarding

Read enough session write-ups and a pattern shows up across the community: high-scoring rounds usually share a few traits, and low-scoring rounds share a different few. None of this is documented by the developer, but the pattern is consistent enough that it is worth adjusting your delivery around.

• Voice contour that follows the reference line (where the pitch goes up and down, and roughly by how much).
• Timing that lands each syllable close to where the reference landed it.
• Clean delivery, meaning sharp consonants, shaped vowels, and a consistent character across the whole clip.

What the judges are not doing is measuring whether your impression was "good," whether it was funny, or whether the room laughed. Those things matter to the table; they do not seem to move the score.

Why impressions that feel right still score low

The most common reason a "great" impression scores badly: you are reacting to the reference instead of mirroring it. Most people's instinct when they hear a clip is to wait, process, maybe laugh, then deliver. By the time you start, the timing has already drifted. The judging seems to care less about how good it sounded and more about how close to the reference you got.

The second-most-common reason: you are softening the consonants. Reference lines often have sharp "t", "k", and "p" hits. Your natural speech softens them. Sharper consonants tend to score higher than the same impression delivered more casually. This is something that shows up consistently in player write-ups even though the developer has not commented on it.

Habits that lift your average fast

Listen for the shape, not the words

Do not try to hear what the line means. Try to hear the shape. Where does the pitch go up, where does it come back down, where does it pause? Sing the line back silently in your head before you speak. Players who hear the shape first almost always score higher than players who try to "perform" the line.

Commit early and do not fix mid-round

Halfway through the round is too late to redirect. You can hear the change in your own voice when you start a syllable and then second-guess it, and the clip does not get better from there. New players constantly restart syllables in their head. Experienced players commit and ride it out.

Match the energy of the reference

A quiet reference line read at full volume tends to score surprisingly low. A whisper delivered as a whisper scores higher than the same whisper delivered as a yell, even when the pitch contour is identical. This is one of the more frequently mentioned observations in the community and it holds up across pack types.

Practical pre-round checklist

Spend ten seconds before each round on these four checks. Most players skip them and lose easy points.

• Mic distance. Roughly a fist's width from your mouth. Closer than that and plosives blow out the input.
• Room tone. If the room got noisier in the last minute, pause and wait for a quieter beat.
• Headphone check. Make sure the reference line is coming through the headphones, not the speakers. Speaker bleed into the mic wrecks the audio.
• Breath. One deep breath before the recording window opens. A flat first syllable is almost always a breath problem, not a pitch problem.

What to skip on round one

Three strategies look smart but tend to hurt your score:

• Adding your own catchphrase at the end. It is fun, but it pushes the actual round past the audio the judges heard and dilutes the per-syllable match.
• Doing a falsetto "shimmer" on vowel-heavy lines. It looks impressive; it tends to score lower than a clean mid-range read.
• Re-doing consonants louder to "compensate." Loud is not clear. Sharpness matters, not volume.

Reading the score screen mid-session

After a few rounds the score screen starts telling a story. If you are consistently scoring 4 to 5 on short, fast lines but tanking on slow lines, the issue is almost always that you are losing timing on the long vowels. If you are scoring well on vowels but poorly on consonants, your mic is probably sitting slightly off-axis and losing the "t" and "k" hits. Treat the scoreboard as a diagnostic, not a verdict.

How pack choice changes your score

Packs with strong, recognizable pitch contours score higher than packs with monotone deliveries. Meme packs with signature "shouts" outscore quiet atmospheric packs, probably because there is more pitch data for the engine to work with. If you are trying to win, load a pack of exaggerated lines. If you are trying to feel like a vocal artist, load a pack with nuance, and accept that the scoreboard will look worse.

FAQ

Does the score reset between modes?

The Studio Mode scoreboard is independent of Dub Mode and Twitch Mode. Each mode uses a different scoring model. Twitch Mode uses chat votes, so the dynamics are entirely different.

Is the scoring fair across accents?

Community discussion suggests the model is broadly accent-agnostic. It compares the shape of your voice to the reference shape, not absolute pitch values. Players with non-American accents generally do not report a penalty in write-ups, though regional vowel differences in some reference clips can feel less natural to mimic. This is observation, not a guarantee.

Can I see my average score after a session?

The desktop build saves per-session stats; the browser build currently only shows per-round scores. If you want a long-term view of your average, the desktop release is the way to go.

Why does Twitch Mode scoring feel so different?

Twitch Mode replaces the judges with chat votes. Chat rewards performance, energy, and entertainment value, not acoustic match. The two formats reward different instincts, so the skills do not transfer one-to-one.

What to take away

You will not get a precise scoring spec from the developer, but the community has converged on a useful rule of thumb. Listen for the shape of the line, commit to the whole delivery, and match the energy of the reference. Pair those with a quick mic check and a pack with strong pitch contours, and the scoreboard climbs within a session.

2026 scoring meta update

A few patterns that have emerged since this guide was first written:

• Consonant sharpness matters more than it used to. Players who record in quiet rooms with USB or headset microphones are pulling away from laptop-mic users by noticeable margins. If you are serious about scores, the cheapest improvement is the audio chain.
• Longer reference clips are scoring harsher. The community has gravitated toward packs with five-to-eight-second clips. Anything over twelve seconds tends to score lower across the board, even on good takes, because the judges seem to drift on longer samples.
• Pitch contour matching is the single biggest predictor. Across the top-scoring players in our test sessions, the trait that correlated most with high scores was the ability to reproduce the pitch swings of the reference, not overall voice similarity. Two takes with identical word delivery but different contour matching usually separate by one to two points.
• Energy at the start of the line matters. Players who hit the first syllable at the same volume and intensity as the reference outscore players who ramp up gradually. A late start costs about a point on average.

None of these are documented by the developer. They are patterns we have noticed across multiple sessions with different packs, hosts, and player skill levels. Take them as a starting point, not a rule.

What to listen for when picking a pack to climb in

If you want to climb the scoreboard and have not picked a pack yet, look for these traits. First, clear pitch movement — if the reference swings more than a third between the lowest and highest syllable, the contour-matching skill pays off. Second, single-voice audio — no backing music, no overlapping characters. Third, captioned lines — captions do not affect the score directly, but players who read the caption usually take a better swing on the second or third try.

Want to put the strategy to work? See Best Voice Packs for archetypes that tend to score well, or How to Build a Voice Pack to design one that suits your voice.