What music mentoring sounds like, and the participants behind the scores
Noise Solution pairs participants with professional musicians. After every session, the participant and their musician record a short reflection video — a moment of looking back together at what just happened. An AI analysis engine turns those reflections into measurable scores on three psychological needs known to support wellbeing. This page shows what those scores actually look and sound like.
Loading snapshot…
The December 2025 scoring change
Around mid-December 2025, the AI's scoring rules tightened. From that point on, no session reached the top of the scale, and average scores dropped. The chart below shows the monthly average for each of the three psychological needs — all three drop together at the same calendar moment.
Any apparent decline you see across the December 2025 boundary is the measurement, not the participants. Where a chart spans that line, the two sides aren't strictly comparable. Where it sits inside one side, the comparison is clean. The smaller notes under individual charts further down flag this again where it matters.
The shape of … sessions
Every dot below is one session, scored by the AI for each of the three psychological needs. Most sessions cluster in the middle of the scale; the spread tells you how much sessions vary on that need. The marked dots are the nine sessions whose words you'll read in the next section. They cluster toward the higher-scoring end of the distribution — selected for the richness of their qualitative material, which tends to be thicker in higher-scoring sessions.
The short vertical line in each row is that need's median session score — the score with half of sessions above and half below. Reading the three ticks left to right: Competence sits at 6, Relatedness around 5.2, Autonomy at 5. Autonomy's median is pulled down by the cluster of sessions sitting exactly at 5; its second, smaller cluster around 7–8 is visible to the right.
Hear what's happening in sessions
Each card below pairs a participant's own words from a session with the AI's measurement of the three psychological needs in that same session. The quote is what they said; the bars are how the AI scored that session for Competence (feeling good at something), Autonomy (feeling in control), and Relatedness (feeling connected to and valued by others).
And where it didn't click
A short counterbalance. For each of the three needs, here's a session where the AI scored low and the participant flagged some friction. Real impact work needs both halves.
Patterns across journeys
The two charts below shift from "what does the population look like" to "what tends to happen as someone keeps coming back." Both are best read as a preview of how this kind of analysis will land once Transceve's rubric has settled and more sessions have accumulated — the shapes you see here are real, but the magnitudes will become more reliable as the data deepens.
What changes between someone's first session and the rest
Each line is one participant who's had more than one session. The left end is their first-session score; the right end is the average of all their later sessions. Lines tilt up where a person's later sessions scored higher than their first; down where the later scored lower.
Worth knowing: a person who started on Transceve's wider rubric and continued through the tightening will show as "later sessions lower" even if their actual experience didn't change. Some of the red here is rubric drift, not decline. Once the rubric stabilises, this same chart will separate within-person change from instrument change much more cleanly.
Does deeper engagement go with higher scores?
Each dot is one participant. Horizontal axis: how many sessions they've had so far. Vertical axis: their average score across those sessions. A line through the cloud shows the rough trend; a flat line means engagement depth and average score don't move together in this sample. The shaded band around the line is the margin of uncertainty for that trend — with around 30 participants per panel, the true direction could plausibly sit anywhere inside the band.
How cohort averages have moved across the seven months
Each panel is one cohort's monthly trajectory across all three psychological needs at once. The first chart cuts the population by age band; the second by sector. With small cohorts (3–10 participants each, drawn from a total of 33), single-month averages will jump around — read the overall direction of each line, not its month-to-month wiggles.
Same caveat as the slope chart applies, with extra force here: a cohort's apparent dip in January isn't necessarily a real dip — it's partly the Transceve rubric tightening at that point. Once the rubric stabilises, these lines will show real cohort movement; for now, treat the chart as a preview of how cohort tracking will work, not as a verdict on which cohorts are doing better.
Each participant, session by session
One card per participant, sorted by number of sessions completed so the most evidence-rich journeys come first. The small chart on each card traces that person's Competence (blue), Autonomy (green) and Relatedness (red) across their own sessions in order. The bars summarise their average across all sessions.
A note on reading these journeys: the sparkline shows raw scores in session order, not a like-for-like comparison across time. Transceve's rubric has been actively tuned during this period (see Methodology), so a dip or rise inside one person's chart partly reflects the engine maturing as well as that person's experience. The shape of the cards — how varied the journeys are between people — is the more reliable read.
How we measure
The scores on this page come from an AI-powered analysis engine called Transceve, which Noise Solution developed to extract structured measurement from the unstructured conversation in each session's reflection video. The framework Transceve uses is Self-Determination Theory, an evidence-based motivational psychology framework that identifies three Basic Psychological Needs underpinning intrinsic motivation and wellbeing.
- Sense of Competence
- Feeling good at something — having capability and growing skill.
- Sense of Autonomy
- A sense of feeling in control — making meaningful choices about one's actions.
- Sense of Relatedness
- A sense of feeling connected to and valued by others.
- The 1–9 scale
- Each Basic Psychological Need is scored from 1 (not at all) to 9 (fully supported).
- Five-pass averaging
- Each reflection video is analysed five times by Transceve and averaged, to manage variability inherent in language-model output.
- Sample shown here
- Sessions delivered between September 2025 and March 2026, across the participants who consented to having their data analysed. We've plotted the data as it is — no correction, no exclusion of earlier sessions — and the chart should be read as suggestive of how this kind of view could be used once Transceve's rubric stabilises.
- What this view doesn't show
- Three things are deliberately left out: cuts by demographic group (the sample is too small to split by gender, age band, or sector without producing misleading per-cell numbers), links to digital-storytelling engagement (post and comment counts aren't in this export), and links to the wellbeing-scale change participants complete at the start and end of their time with us (that data is collected separately and will combine with this view in a later project phase). Each was considered and held back rather than shown thinly.
- The rubric is still emergent
- Transceve has been actively tuned during this period — the scoring rubric tightened in mid-December 2025, and the way scores are produced may continue to evolve. As a result, a "5" in October and a "5" in March are not strictly the same measurement. We're showing the page in this form so the visualisation choices can be evaluated for shape and storytelling; precise scale-comparable measurement will follow once the underlying rubric has settled.
- Why the per-person journeys are real signal
- We checked: scores swing meaningfully both between participants (some have higher baselines than others) and within a single participant across their own sessions, so the journeys in the previous section reflect genuine session-to-session change rather than noise around a fixed setting.
- A note on the score and the experience
- The score and the lived experience are not the same thing. We have learned, by checking the AI's scores against the participants' own ratings of each session, that the AI's scoring scale is narrower than the participants' (the AI tends to use the middle of the scale, while participants sometimes use the very top). When you read a quote alongside a moderate-looking score, the participant's own experience may have been more positive than the number alone suggests. Pairing the words and the number is deliberate — it shows both at once.