CONCEPT ILLUSTRATION
FEATURED / SOUND-BY-SOUND PRACTICE
Articulation Trainer
Practice a word. See the sounds. Bring the individual parts of an attempt into view, with feedback that connects what you produced to the task you are trying to complete.
SPEECH PRACTICE, MADE MEASURABLE
Speech is a motor skill.
Train the control system.
Make timing, sound transitions, and vocal control visible. FluentPlay connects speech analysis with voice-driven games, so every attempt has a target—and feedback you can use.
7 practice games2 analysis toolsBuilt by someone who stutters
A SPEECH ATTEMPT, UNFOLDEDOne transition. Many moving parts.
Concept visualization01 / HOLDmmm
03 / LANDooo
Look between the sounds—not just at whether the word was completed.
01
Analyze the attempt.See the speech signal.
02
Practice the movement.Give the task a purpose.
03
Review what changes.Keep the context.
01THE APPROACH
The message is there.
Practice the movement.
Knowing what you want to say and producing it are not the same experience. FluentPlay starts with that distinction: practice should offer more than a judgment about whether a word came out.
“The sentence is clearly formed ahead of my execution attempt.”Will Carbone
Founder, FluentPlay · A personal account of stuttering
01 / INTENTION
Meaning
What you want to communicate.
02 / FORMULATION
Language
The words and structure of the message.
OUR PRACTICE FOCUS03 / PRODUCTION
Movement over time
The onset, timing, transitions, and vocal output that a practice task can make visible.
A simplified training perspective—not a claim that stuttering has one cause or that every sentence is planned in advance.
02THE FEEDBACK LOOP
Not just another attempt.
A more informative one.
The game gives the attempt a purpose. Measurement makes a difference visible. The next attempt gives you somewhere to apply it.
The transition lab INTERACTIVE WEBSITE DEMO
Synthetic data · no microphonemmm → ooo
ReferenceAttempt
+80ms
180ms
2.40s
0 ms+300 ms
Connected360 ms
Move either slider. The trace changes—not your speech. These controls explain feedback; they do not simulate a treatment response.
Give the attempt a target.
A game makes one aspect of speech relevant to the task. Here, the illustration focuses on maintaining voicing between two sustained sounds—not on grading an entire conversation.
The reference is specific to this sustained-voicing illustration. Natural speech includes intended silences and unvoiced sounds; a gap is not, by itself, a stuttering diagnosis.
| Attempt | Start offset | Voicing gap |
|---|
03THE MOTOR-LEARNING IDEA
The musician’s principle.
Applied to speech.
A musical phrase is more than a sequence of correct notes. Timing, transitions, and the relationship between movement and sound matter. FluentPlay borrows that practice logic—not a claim that stuttering is a lack of practice.
Research on musical training and models of speech control motivate the same question: can carefully structured feedback help make a practiced movement more reliable?[1][3]
What the cerebellum research actually shows
Hutchinson and colleagues studied professional keyboard players, not specifically violinists. Male musicians had greater cerebellar volume than male controls, and volume correlated with practice intensity; the female comparison was not significant. This was an association, not proof that larger volume causes better control—and not evidence that FluentPlay changes the cerebellum. Read the study ↗
DIFFERENT INSTRUMENTS / A RELATED PRACTICE LOGIC
MUSICIntend→Play→Listen / adjust
SPEECHIntend→Speak→Observe / adjust
An analogy for practice design. Not a neurological equivalence.
04THE PRACTICE COLLECTION
Different challenges.
One learning loop.
Seven voice-driven games. Each gives a specific part of the attempt a job to do. Explore the practice targets below.
CONCEPT ILLUSTRATION
SOUND TRANSITIONS
Sound Bridge
Practice the movement between two sounds. Hold, transition, and land—with the bridge itself in focus.
CONCEPT ILLUSTRATION
CHOSEN WORDS & PHRASES
Summit
Turn a chosen speech target into an ascent. Learn the production the task calls for, then return to it through play.
CONCEPT ILLUSTRATION
PHRASE-LEVEL CONTINUITY
Rainbow Syllables
Carry a phrase across its syllables. Keep the full sequence in view instead of reducing speech to one word.
CONCEPT ILLUSTRATION
VOCAL LEVEL & PRECISION
Rhythm Pad
Make “more” and “less” specific. Use voice level to approach a target rather than simply trying to be louder.
CONCEPT ILLUSTRATION
RHYTHM & SEQUENCE TIMING
Cadence
Give speech timing a visible structure. Practice pace and rhythm as defined tasks, not vague instructions.
CONCEPT ILLUSTRATION
VOCAL TARGETING
Bubble Hunt
Bring vocal control into an active scene. Match the task’s target while keeping track of what your voice is doing.
These are illustrated product overviews, not embedded game builds. A completed game target is a task outcome—not a clinical fluency score.
See the analysis tools05THE ANALYSIS TOOLS
ANALYZE + PRACTICEThe attempt has
The attempt has
a record.
The games structure practice. OpenMic and Two-way Conversation bring speech into view at different levels of structure. The goal is a useful connection between what happened and what to practice next.
OpenMic Analysis console
Review speech at session, word, syllable, and phoneme levels. Scripted and free-speech modes provide different contexts for examining the signal.
Two-way Conversation In development
Extend the analysis view into back-and-forth speech, where the next utterance is less predictable. Keep conversation context alongside the signal.
OpenMic / analysis workspaceIllustrative
PREPARED TEXT / SELECT A WORD TO INSPECT THE SAMPLE
SELECTED WORDspeak
DETECTED ONSETS2
VOICED DURATION0.43 s
REVIEW MARKERInspect
RECORD TYPESynthetic
A prepared phrase gives the attempt a known language target. Select a word to see how a signal record can be inspected without reducing it to a single score.
INTERFACE CONCEPT · The words and values are synthetic. An onset count does not diagnose a block or reveal its cause.
06PAD / THE MEASUREMENT FRAMEWORK
A profile.
Not a verdict.
PAD brings acoustic features and speech-recognition output into a shared measurement framework. Its purpose is to keep the changing attempt visible—not to turn a person’s voice into a pass or fail.
Compare the shape of different sessions without assuming that every axis should rise. Speaking faster, for example, is not automatically speaking better.
01Keep the feature-level record.
02Compare like tasks and conditions.
03Keep personal experience in view.
Audio features are not direct measurements of anxiety, muscle tension, air pressure, or brain activity. Those interpretations need separate evidence.
Session geometrySynthetic profiles
DRAG HORIZONTALLY TO ROTATE · OR USE THE ROTATE CONTROL
Display-only illustration, not the PAD scoring algorithm. The drawing normalizes axes for layout; raw sample values are shown above. “Acoustic pressure” is an audio-derived proxy, not physical pressure. No overall score is calculated.
07THE SCIENTIFIC FOUNDATION
A clear hypothesis.
An honest boundary.
Motor-learning research informs the design. Whether this particular training approach leads to durable, useful changes in speech remains a question to test.
01 / LONGITUDINAL RESEARCH
Training and structural change.
Hyde and colleagues observed brain-structure changes over 15 months of instrumental training in children, alongside gains in relevant motor and auditory skills. This supports studying experience-dependent change; it does not establish a speech-treatment effect.
Hyde et al. · Journal of Neuroscience, 2009 ↗02 / CROSS-SECTIONAL RESEARCH
Cerebellar differences in musicians.
The often-cited volume finding concerned professional keyboard players and was significant in the male comparison, not the female comparison. Correlation with practice is informative, but does not show that increasing volume improves speech.
Hutchinson et al. · Cerebral Cortex, 2003 ↗03 / COMPUTATIONAL THEORY
Prediction alongside feedback.
The DIVA framework describes learned feedforward commands alongside auditory and somatosensory feedback. It provides a model for asking how sensory errors can inform future speech movements—not validation of FluentPlay itself.
Guenther & Vladusich · Journal of Neurolinguistics, 2012 ↗04 / EXPERIMENTAL SPEECH RESEARCH
Learning has to be measured.
Kim and colleagues found reduced speech auditory-motor learning in children and adults who stutter under the tested conditions. Such findings make the learning mechanism a research target; they do not imply that more repetition alone will solve it.
Kim et al. · Neuroscience, 2020 ↗Is the measurement useful?
Test agreement with independently reviewed speech, sensitivity to recording conditions, and the relationship between the signal and the speaker’s experience.
Does learning remain?
Separate success during a game from performance on later attempts without the same cues. Retention is an outcome to measure—not something a moving score can establish.
Does it transfer?
Test unpracticed material, different contexts, and everyday communication. Include effort, comfort, and participation, rather than relying only on game performance.
PROPOSED VALIDATION QUESTIONS. This preview does not report a completed trial, clinical effect size, participant results, or research approval.
BEHAVIORAL LAYER
Audio & task performance
Develop the relationship between speech features, task demands, and personal reports before making claims about hidden physiological states.
Product developmentMULTIMODAL DIRECTION
EEG, EMG & respiration
Explore synchronized recordings around speech attempts. Planned physiological research is separate from what the browser tools currently measure.
Research directionPLANNED CAPABILITY
fNIRS research
Build toward speech studies with fNIRS recordings. Hardware acquisition, study readiness, and validation are not presented here as completed work.
Planned acquisitionThis is a development roadmap, not a participant-recruitment notice. No active study, named research partner, ethics approval, or owned fNIRS system is implied.
What FluentPlay is proposing
Structured practice with interpretable feedback may help people develop more reliable speech control. The platform is being built to investigate that possibility. It does not currently claim to diagnose stuttering, repair a brain structure, or provide a clinically validated treatment.
08MADE FOR PEOPLE
One platform.
Different perspectives.
Practice on your terms.
Keep your own voice.
Choose a target that matters to you. Explore feedback without treating a score as a measure of your worth—or as the whole story of how speaking feels.
Explore the practice collectionBUILT BY SOMEONE WHO STUTTERS
Will Carbone
Founder · FluentPlay Technologies
“Self-acceptance can coexist
with ambition for change.”
with ambition for change.”
FluentPlay grows out of that belief. A person can value their voice and still want better tools for practicing speech. The work is to build those tools with measurable feedback, clear limits, and respect for the experience that a microphone alone cannot capture.
Contact Will09A FEW CLEAR ANSWERS
Before you
go further.
Is FluentPlay a clinically validated treatment?
Not on the evidence presented here. FluentPlay is a measurement and practice platform in development. Research on motor learning informs the idea; the effectiveness of the actual platform must be tested separately. No cure or guaranteed fluency outcome is offered.
Does this website preview listen to me?
No. This standalone file never requests microphone permission, records audio, calls a speech-recognition service, or uploads data. The interactive displays use synthetic examples. Opening an external FluentPlay app leaves this preview; that app has its own behavior and data requirements.
Are the game scenes and PAD profiles real results?
No. The scenes are concept illustrations. The lab traces, word records, and profile values are synthetic and are labeled as such. The profile drawing is not the PAD algorithm, and none of the examples represents a person’s improvement.
Why not just count fluent and stuttered words?
That outcome can be useful, but it is not the whole design target here. FluentPlay’s approach is to retain more of the attempt—its timing, transitions, and observable signal—so a practice task can focus on a specific feature. A richer display still needs measurement validation.