Choicer Voicer
All guides

Scoring guide

How Choicer Voicer Scoring Works

The score compares the shape of your take with the original, not the identity or absolute range of your voice.

Use the three separate scores to find your weakest dimension, then change one thing on the next take. The result is for entertainment and practice.

Record a take

Step by step

  1. 01

    Choose a focused scene

    A 2 to 15 second segment gives the browser enough sound to compare while keeping the performance repeatable.

  2. 02

    Listen for the shape

    Notice where the original rises, falls, gets louder, softens, starts, and pauses. The score follows those patterns rather than the words themselves.

  3. 03

    Record a clean take

    Use headphones when following the original and record in a quiet room. Background audio can blur pitch and energy analysis.

  4. 04

    Read each dimension

    Pitch contour, rhythm and energy, and timing are shown separately. The post-take waveform comparison also shows timing alignment at a glance. The weakest score usually gives the most useful next target.

  5. 05

    Retake one change at a time

    Keep the scene, microphone, and room consistent. That makes your own attempts more meaningful to compare.

More detail

Pitch contour

The browser tracks voiced pitch over time, removes the median level from both performances, and compares the remaining rise-and-fall patterns.

Because the middle pitch is removed first, a higher or lower voice can still match the contour. Absolute vocal range is not the target.

Rhythm and energy

The energy shape is normalized before comparison, so the score focuses on changes in emphasis instead of raw microphone volume.

A small timing search can absorb minor capture delay. It does not rescue a performance whose stresses and pauses follow a different pattern.

Timing and voice coverage

Timing compares where both tracks contain voiced sound and where both are quiet. Starting late, filling a pause, or finishing too soon can lower it.

The total also applies a voice coverage adjustment. Speaking far less or far more than the reference can reduce the overall score even when one dimension looks strong.

How the total is combined

The current calculation gives 40 percent to pitch contour, 30 percent to rhythm and energy, and 15 percent to timing, plus a 15 percent baseline before the voice coverage adjustment.

Scores are most useful for comparing your own retakes of the same scene. Different scenes, microphones, and rooms do not form a universal leaderboard.

Six practical ways to improve a retake

Wear headphones — speaker audio leaking into the microphone makes the browser hear both performances and adds noise to the contour comparison.

Match the emotion, not the absolute pitch — follow where the character rises and falls without forcing your voice into an unnatural register.

Start and stop with the character — timing compares voiced and quiet windows, so an early entrance or a long trailing ending adds mismatches.

Use the 50 ms timing nudge — a small offset can correct consistent capture delay when the take sounds right but appears slightly misaligned.

Keep your microphone distance steady — moving during the take changes the energy curve independently of the performance.

Trim the clip tighter — one compact line between 2 and 15 seconds gives you a clearer, more repeatable target than a long scene with quiet gaps.