You're coming. can be a statement, a question or an expression of surprise in the right context. The words alone do not reveal that difference, and when listening to yourself it can be hard to know what to compare.
Pitch visualization displays how pitch changes over time, making rises, falls and lengthening easier to inspect. It can support feedback, but it cannot reduce natural intonation to one perfect line.
What the curve can—and cannot—show
A pitch track primarily reflects changes in vocal-fold vibration frequency. It is not the whole of intonation, much less a naturalness score. Loudness, duration, prominence, pauses, voice quality and context also shape what a listener hears.
Levis and Pickering discussed speech visualization for teaching discourse intonation in 2004. This offers a teaching route, not proof that copying a picture teaches every communicative tone. Speakers, emotions, accents and sentence types naturally produce different contours.
You can start without specialist software
Praat is a widely used free speech-analysis tool that can display recordings and pitch tracks. An ordinary recorder also supports the core loop: record, listen and compare. The curve is optional, not a prerequisite.
Choose a sentence with a clear purpose, such as confirming an arrangement: So we meet at three? Prepare a natural model and record your version.
1. Listen before looking: is it confirmation, challenge or a statement? Write down the communicative purpose. 2. Inspect one feature: compare only the ending, lengthening of a keyword or a pause. Do not chase the whole graph. 3. Hide the curve and imitate: record three versions and select the most natural meaning, not the closest visual overlap. 4. Change the content: say So we meet on Friday? without the graph and check whether the confirming intention survives.
A useful beginner contrast
The shop closes at six. and The shop closes at six? share words but can convey different intentions. Hear the contrast first, then see whether the final pitch movement differs.
Do not treat “questions rise, statements fall” as an exceptionless rule. Intonation depends on information focus, attitude, context and variety. A small bump in a track may also be a measurement artefact, not something to correct.
Three ways visualization misleads
Assuming more overlap is always better. Speakers have different pitch ranges, and visual identity is not necessary for effective communication.
Fixing only the final note. Listeners may care more about which word is prominent or where a thought group ends. Longer sentences also need thought-group and stress practice.
Looking without listening. In real conversation there is no graph. Every visual inspection must return to blind listening and independent speech.
MirrorSay provides a model, not an invented score
Say your intended meaning to MirrorSay, listen to a natural formulation and record your repetition for comparison. If you want a visual check, inspect your recording and the model in an external speech tool. MirrorSay itself does not provide a pitch curve or an intonation score.
Practise one communicatively meaningful contrast at a time, then change the words and speak without the display. If you cannot yet hear the difference, prioritise perception; if you are imitating every small detail, revisit your pronunciation priorities.
References
- Levis, J., & Pickering, L. (2004). Teaching intonation in discourse using speech visualization technology. *System*, 32(4), 505–524. DOI
- Tan, M. (2021). Study Intonation: A mobile-assisted pronunciation training application. *TESL-EJ*, 25(3). Open review
- Praat: official speech-analysis software.
*MirrorSay is a free AI speaking-practice app: say a sentence, see a natural way to express it, then shadow the model. No sign-up, and speech is processed on-device whenever possible.*