Science & Trust
Trust matters more than cleverness. Melloba is designed to describe everyday interactions responsibly, and to be transparent about the limits of that description.
What Melloba measures
How the conversation flowed — who talked, the turns, the replies — comes from who spoke when, not from what the words mean. Learning Ability, Thinking Skills and Curiosity come from the words themselves, so they need the words to be heard clearly and are not available for every recording.
From who spoke when
From the words
The conversation timeline
Left to right is time through the recording. Each lane shows one thing, on its own scale, starting from zero. Where nothing was measured, the lane is left empty, and each reply is its own dot — never joined up. The first view carries no numbers: the shapes tell the story, and the detail is one tap away.
The coral is your child's share of everything said in this recording, and the teal is yours. There's no right split — it's common for a young child to say less than the grown-up, and what matters more is that the turns kept going.
A turn is one of you speaking and the other answering within a few seconds. The shape rises where turns came thick and fast, and settles where they slowed down.
Each dot is a time your child answered you. The lower a dot sits, the sooner your child began to reply after you finished. A slower reply isn't a weaker one — it can simply mean your child was thinking.
It rises when there was plenty of talking and it was shared evenly between you. It's worked out from who spoke when, rather than measured separately.
You did more of the talking than your child.
Two voices
Melloba's own analysis of the sound works out who spoke when. The ring shows how the talking was shared across the whole recording: your child's share in coral, starting at twelve o'clock, and yours in teal completing it.
The two arcs always make the whole circle between them. There is no track, no needle and no number — it isn't a gauge, and there is no right split.
Learning Ability, Thinking Skills, Curiosity
Learning Ability, Thinking Skills and Curiosity each appear on their own — three petals from one base, never added up into a score. Each has its own colour, which says which dimension it is, never how well.
Illustration of one conversation. Each petal's length is that dimension's own level, compared only with your child's earlier recordings.
Levels
Learning Ability, Thinking Skills and Curiosity each stand on their own. Once a dimension has been measured in at least three of your child's earlier recordings, it can show a level — Smart, Growing or Emerging — describing how strongly it came through in this conversation compared with your own child's earlier ones. A level always belongs to its dimension. There is no overall level, and levels are never compared with other children.
Learning Ability, Thinking Skills, and Curiosity are interaction-based insights from the recorded conversation. They are not measures of intelligence or clinical assessments.
Learning Ability was quieter than usual in this chat.
Learning Ability came through much as usual in this chat.
Learning Ability came through more than usual in this chat.
The petal's length shows the level; its colour never changes. Compared with your own earlier chats — never with other children.
What we don't claim
Melloba describes what happens in an interaction. It does not diagnose conditions, assess whether a child is developing "correctly", or label a child as ahead or behind. It never gives your child an overall score or an intelligence rating, and it never ranks children against each other.
Melloba is not a medical device and is not a substitute for professional advice. Nothing in the app should be read as a clinical assessment. If you have concerns about your child's development or health, please speak with a qualified professional.
Melloba does not predict future outcomes or risk. It reflects patterns in the interactions you record — nothing more. We deliberately avoid predictive claims because they are easy to misread and hard to justify responsibly.
A short recording is a small window. Audio quality, context, mood, and countless everyday factors all shape what is captured. Each report explains, one tap away, how it was measured and what it cannot tell you, rather than presenting numbers as certainties. Learning Ability, Thinking Skills and Curiosity also need the words to be heard clearly, so they are not available for every recording.
How AI is used
Melloba uses AI to turn the recording into text. Who spoke when is worked out by Melloba's own analysis of the sound. Every measurement and every level is then calculated by Melloba's own fixed rules, so the same recording always gives the same result. How the conversation flowed — who talked, the turns, the replies — comes from who spoke when, not from what the words mean. Levels are calculated, not decided by AI.
Privacy
When you set up your child, you confirm that you are their parent or guardian and that you consent to Melloba recording and processing these moments.
The audio file is deleted once Melloba has finished with it. What stays is the report and the results behind it, until you delete your account.
Nothing is shared for research unless you choose to allow it. That option is off by default.
Melloba's measures build on ones used in child-language research. The combination is Melloba's own and is not a validated clinical instrument.