Description: A karaoke score of 90, pitch 36/40, stability around 24/30, and perfect marks for long notes and dynamics. Why can those results coexist, and what should you test next?
The singing sounded fine. The display kept showing strong pitch-match sections. The long-note and dynamics categories even hit their maximum. Then the results screen delivered its verdict: pitch 36/40, stability approximately 24/30. Again. Stability had apparently leased its own permanent parking space at 24.
This is the story of a very specific karaoke problem: trying harder to follow a pitch display, reaching 90 overall, and still being unable to move one category.
What do the five JOYSOUND categories mean?
In Analysis Scoring Master, the displayed category maxima are pitch 40, stability 30, dynamics 15, long notes 10, and technique 5, making 100 in total. The observed result was 36 for pitch, about 24 for stability, 15 for dynamics, and 10 for long notes. The exact technique score was not recorded. If all five added to the observed overall 90, technique would account for the remaining five points, but that is an inference, not a copied result screen.
JOYSOUND describes pitch as a score based on matching the guide melody across the whole song. Stability measures unwanted variation in pitch. Dynamics rewards changes in expression through different song sections. Long notes assess whether sustained notes remain at the right pitch and steady. Technique covers scoops, ornamentation, and vibrato.
So the big contradiction is only apparent: a perfect long-note score is not a perfect stability score.
Why didn't all those high pitch indicators turn into 40/40?
The live graph reports pitch agreement section by section. The result screen turns the entire song's pitch agreement into a category score. A series of excellent moments does not erase weaker sections elsewhere.
Earlier performances had shown an overall agreement around 83%, even while some short sections approached 90%. Later, the pitch category reached 36 out of 40. Those numbers describe different things. Dividing 36 by 40 and announcing a certified 90% pitch-match rate would invent a conversion rule JOYSOUND has not supplied.
The official advice is to follow the main melody. Copying a professional singer's expressive bends does not automatically guarantee a higher machine score, even when the song sounds natural to listeners.
The trap: correcting every note while looking at the bar
One clue came from the singer's own observation: “I keep adjusting toward the bar in the middle of the note.” If a note starts slightly low, it gets pushed up; if it overshoots, it gets pulled down. That repeated steering could introduce extra movement. It is a plausible explanation for reduced stability, not a confirmed diagnosis.
The next idea was to sing like a robot, keeping notes perfectly level. But neither deliberate rigidity nor relaxing back into natural singing broke the approximate 24-point pattern. The stability judge continued to act like an extremely polite bouncer: “Lovely sustained note. Still 24.”
There may be trouble at note attacks or transitions, but the score alone cannot identify the exact moments.
A useful experiment instead of more guessing
Use the same song, key, microphone setup and approximate level for three runs. First, sing while actively correcting to the visual pitch bar. Second, sing from memory without chasing the display. Third, look ahead to the next target note but avoid continuous mid-note correction.
Record pitch, stability, dynamics and long-note scores each time. If stability moves from 24 toward 26, that is evidence that the changed approach may matter. If it stays at 24, inspect the post-song pitch trace and, where available, saved playback before choosing the next adjustment.
JOYSOUND's official Master guide says the results screen can show the full-song pitch trace. On JOYSOUND MAX GO, MAX2 and MAX, Pitch Training Mode selects sections below 85% agreement for repeated practice. Availability depends on the machine and feature.
So how do you get past 24?
Don't spend the next session trying to make already-perfect long notes even longer. Focus on starting notes cleanly, moving between notes without excessive steering, and comparing controlled attempts. That's a better hypothesis test than declaring that every singer must become a robot.
The audience says, “That sounded good.” The machine says, “Stability: 24.” After giving full marks for long notes, the machine ought to offer an organizational chart of its judging department.
References (2)
- JOYSOUND, Analysis Scoring Master: how it works: https://www.joysound.com/web/s/function/bunsekimaster
- JOYSOUND, Scoring and ranking: official FAQ: https://www.joysound.com/web/s/help/ranking/analysis
