Disclaimer

SpeechTherapyMagic is an educational practice tool

SpeechTherapyMagic is provided by Digital Mundo, LLC for educational, informational, and supplemental practice purposes only. It is a library of practice material and a set of tools built around it: games, interactive stories, worksheets, flashcards, word lists, articles, recording and pronunciation-scoring features, progress charts, and downloadable resources. Everything on the platform exists to give a child more repetitions of a target sound, and to give the adult working with them something to organize and track that practice.

That is the whole of what it is. Practice is one part of speech and language work; it is not the part that decides what to practise, why, or whether the plan is working. Those decisions belong to a qualified person who has met the child.

The Service is designed to be used alongside professional care, schoolwork, or a home practice routine an adult has set up — not in place of any of them.

Not medical or clinical advice, diagnosis, or treatment

The Service does not diagnose, evaluate, treat, cure, or prevent any speech, language, communication, cognitive, feeding, swallowing, hearing, voice, fluency, learning, developmental, medical, or other condition. Using the Service does not create a therapist-client, speech-language pathologist-client, or healthcare provider-patient relationship with Digital Mundo, LLC or with anyone who contributes to it.

Nothing on SpeechTherapyMagic — no article, score, chart, recommendation, activity result, or piece of generated feedback — is medical advice, clinical advice, therapy, or a professional opinion about any individual.

The Service does not:

• Diagnose. It does not identify, rule out, or name a speech sound disorder, language disorder, apraxia, fluency disorder, voice disorder, hearing loss, developmental delay, learning disability, or any other condition.
• Screen. A low score is not a screening result and a high score is not a clearance. The platform is not a screening instrument and was not designed, validated, or normed as one.
• Evaluate. It does not produce an assessment, a severity rating, an eligibility determination, or anything that should be entered into a report as a test result.
• Treat. It does not deliver speech-language therapy, set therapy goals, or replace a treatment plan. Practice material is not a course of treatment.

Using the Service does not create a therapist-client, speech-language-pathologist-client, clinician-patient, healthcare-provider-patient, teacher-student, fiduciary, or other professional relationship with Digital Mundo, LLC or with anyone who works on it. We do not know your child, we do not review your recordings, and we are not in a position to advise you about them.

Not a medical device, and not a standardized assessment

SpeechTherapyMagic is not a medical device. It has not been cleared, approved, registered, or otherwise reviewed by the U.S. Food and Drug Administration or by any comparable regulator in another country, and it is not intended to be used to diagnose, cure, mitigate, treat, or prevent any disease or condition.

It is also not a standardized or norm-referenced assessment instrument. Standardized tests are built to a specific protocol, administered under controlled conditions, and scored against a normative sample collected for that purpose. Our scores are none of those things: there is no normative sample behind them, no standard administration, no scaled score, no percentile, and no age equivalent. A number produced here cannot be compared to a number produced by a published test, and it should never be reported as though it were one.

Practically, that means scores from this platform do not belong in an eligibility decision, a diagnostic report, or an IEP present-levels statement as test data. They can reasonably be described as practice data — how much practice happened, on which targets, and which way the trend is moving — which is a genuinely useful thing, and a different thing.

How automated pronunciation scoring works, and where it fails

Some features record a child saying a word and return a score, a per-sound breakdown, or an error flag. This is done by machine learning models, not by a person. No human listens to a recording before feedback appears on screen.

These models are probabilistic. They estimate how closely a production matched an expected pronunciation, and they are wrong in both directions:

• Missed errors. A production that a clinician would mark as incorrect can be scored as correct. A passing score is not evidence that a sound is accurate.
• False flags.A correct production can be marked wrong. A failing score is not evidence that anything is wrong with a child's speech.

Accuracy varies with conditions that have nothing to do with the child's speech: microphone quality, background noise, how far the child is from the device, room echo, dropped audio, and network conditions all move the numbers. It also varies with the speaker — young children, accented and dialectal speech, atypical voice quality, very quiet or very loud talkers, and children with significant intelligibility differences are all harder for these models than a clear adult speaker reading a word list. Automatic speech recognition is trained predominantly on adult speech, and child speech remains the harder problem in the field.

Text-to-speech audio, generated practice sentences, generated stories, and other automated content carry the same caveat. Safeguards and moderation are in place, but automated material is not individually reviewed by a person before you see it. Review anything generated before you rely on it or share it with a child, student, client, or family.

No guaranteed outcomes

No specific outcome is promised or guaranteed. Digital Mundo, LLC does not guarantee speech, language, academic, therapeutic, behavioral, or developmental improvement, or that any score, recommendation, activity, or generated material will be accurate, complete, or suitable for a particular child.

No particular result is promised. Digital Mundo, LLC does not guarantee speech, language, academic, behavioral, or developmental improvement, does not guarantee that any score, activity, recommendation, or generated material will be accurate or appropriate for a particular person, and does not guarantee that practice on this platform will shorten, replace, or reduce the need for professional services.

Progress in speech and language work depends on the individual child, the underlying reason for the difficulty, the quality and consistency of professional support, and a great many things outside any product's control.

Adult supervision and individual suitability

A parent, caregiver, educator, or licensed speech-language pathologist is responsible for supervising use as appropriate to the child's age and needs, and for deciding whether a given activity, prompt, recording feature, word list, or download is suitable for that individual. We publish general-purpose material; we cannot know which of it fits a specific child.

Qualified professionals remain responsible for evaluation, clinical judgment, goal selection, treatment planning, accommodations, documentation, and care decisions. Where the platform offers a suggestion — a next target, a recommended activity, a difficulty level — it is a convenience based on activity history, not clinical reasoning, and a professional's judgment supersedes it.

Stop any activity that causes a child distress, discomfort, frustration beyond the productive kind, or any safety concern, and raise it with the professional supporting them.

Developmental information, norms, and articles on this site

This site publishes general educational content about speech and language development: ages at which sounds are typically acquired, speech sound development charts, articles on phonological processes, descriptions of conditions, and similar reference material. It is written to be useful to parents, teachers, and students, and it is general information about populations — never guidance about a specific child.

Developmental norms in particular are easy to misread. They describe the age by which a large proportion of children in a studied sample produced a sound; different studies use different criteria and arrive at different ages for the same sound, and the spread of normal variation around any of those ages is wide. A child who is behind a chart may be developing typically. A child who is on schedule for every sound may still have a difficulty worth looking at. Charts do not account for hearing history, language exposure, bilingualism, dialect, oral-motor factors, or anything else about a particular child.

Use this material to understand the landscape and to ask better questions. Do not use it to decide whether a child needs help, to decide that they do not, or to postpone an evaluation. If something about a child's speech worries you, or worries their teacher, a referral to a licensed speech-language pathologist is the appropriate next step regardless of what a chart says.

ArticScore, published results, and the developer API

ArticScore is our pronunciation-scoring engine, offered both inside SpeechTherapyMagic and as an early-access developer API. Performance figures published anywhere on this site — accuracy, correlation with human raters, comparisons with other engines — come from our own internal testing on limited datasets at modest sample sizes. They have not been independently verified or third-party certified, and we do not guarantee that they will hold on your audio, your speakers, or your recording conditions.

Where a comparison with another product appears, it reports what we measured on one specific sample under one specific setup. It is not a general ranking of the products and should not be read as one. The method, datasets, sample sizes, and confidence intervals are published in full on the ArticScore benchmark page, which is the right place to check the numbers rather than restating them here.

The engine is in active development. Scores, thresholds, phoneme outputs, and model behaviour change as it improves, which means results are not guaranteed to be stable across versions. The developer API is early access: endpoints, response fields, limits, and terms may change, and it should not be treated as a frozen interface. Integrators should read the API documentation and the access terms, and should design their own products on the assumption that a score is an estimate.

If you build a product on ArticScore, the obligations that attach to your end users — notice, consent, supervision, and any clinical or educational claims you make — are yours, not ours. We process audio on your behalf; we do not stand between you and your users.

Third-party services, subprocessors, and external links

Parts of the Service depend on third parties. Pronunciation scoring is done by ArticScore, our own engine, on our own servers, but generated content, email delivery, error monitoring, analytics, authentication, hosting, and payments all involve outside providers. Each of those providers operates under its own terms and privacy policy in addition to ours.

The providers who process data on our behalf are named on our subprocessor list, and the terms that govern that processing are in our data processing agreement. If a third party's service is degraded or unavailable, features that depend on it can fail or return worse results, and that is outside our control.

This site also links to outside websites — research papers, government and regulatory sources, professional associations, and other products. Those links are provided for reference and are not endorsements. We do not control the content, accuracy, availability, or practices of any site we link to, and a link does not imply that we agree with what is on the other end of it or that any relationship exists.

Trademarks and product comparisons

Product names, brand names, logos, and trademarks referenced on this site belong to their respective owners. We use them only to identify the products being discussed — in comparison articles, alternatives pages, integration documentation, and technical write-ups — which is nominative reference, not a claim on anyone's mark.

No affiliation, sponsorship, partnership, endorsement, or approval by any of those companies is claimed or implied, in either direction. Comparison pages reflect our own reading of publicly available information and our own testing at a point in time; competing products change, and the responsible thing is to verify current details with the vendor before making a purchasing decision. If you believe we have described your product inaccurately, tell us and we will correct it.

Testimonials and individual experiences

Quotes, reviews, case descriptions, and success stories that appear on this site describe the experience of the individual who gave them. They are not typical results, not average results, and not a prediction of what any other family, clinician, or classroom will experience. No compensation is required for a statement to be disclosed, and where a person quoted has a relationship with the company — an employee, a co-creator, an advisor, or a paid partner — that relationship is stated alongside the quote rather than left for the reader to infer.

Availability, changes, and “as is” provision

The Service is provided on an “as is” and “as available” basis. We do not warrant that it will be uninterrupted, timely, secure, or error-free, and we do not offer an uptime guarantee. Maintenance, outages at providers we depend on, bugs, and capacity limits all happen.

Features, content, scoring behaviour, activities, and pricing change over time. Material may be added, revised, or removed, and content that is accurate when published can become out of date. We do not commit to keeping any particular feature, activity, or resource available.

The legal terms that actually govern your use of the Service — the warranty disclaimers, limitation of liability, subscription and cancellation terms, acceptable use rules, dispute resolution, and everything else of that kind — are in the Terms of Service. They are stated there rather than restated here, and this page is subject to them.

Clinical, school, and organizational use

SpeechTherapyMagic is not a healthcare provider, does not bill or process insurance, and does not function as an electronic health record or a medical billing system. We describe our own data practices and put them in writing; we do not claim certification under HIPAA, FERPA, or COPPA, and for COPPA no such certification exists in the first place.

Clinicians, practices, schools, districts, and other organizations are responsible for their own compliance obligations, including any privacy impact assessment, records requirement, consent process, procurement review, or supervision rule that applies to them. Consult your own compliance requirements and counsel before using recording features, storing student information, or deploying the platform to a caseload or a classroom. Where an organization needs written data terms, our data processing agreement and subprocessor list are published, and school-specific provisions are addressed in the Terms of Service.

Licensed professionals remain bound by their own licensure, scope of practice, and professional ethics. Nothing here alters those obligations, and no feature of this platform should be used in a way that conflicts with them.

Recordings, privacy, and children's data

How audio is handled — what is recorded, what is kept, for how long, under whose consent, who processes it, and how to delete it — is documented in full elsewhere, and those pages are the authoritative statement rather than anything summarized here.

• Privacy Policy — what personal information the platform collects, how it is used, and the choices available to you.
• Children's privacy and voice recordings — how COPPA treats a child's voice, and exactly what happens to a recording on both the app and the developer API.
• Data processing agreement and subprocessor list — the written terms and the named providers.

Children do not create accounts on SpeechTherapyMagic; an adult holds the account and the child practises within it. Consent for anything involving a child's recordings runs through that adult.

Emergencies and urgent concerns

The Service is not designed for emergencies and is not monitored for them. Nobody is watching the platform for signs that a child needs help, and messages sent to us are not an appropriate channel for an urgent situation.

If you believe someone may be in danger or experiencing a medical emergency, call 911 or your local emergency services immediately. A sudden change in a child's speech, voice, swallowing, or alertness — particularly a loss of a skill they previously had — is a reason to contact a physician promptly rather than to practise.

Questions, corrections, and contact

If something on this page is unclear, or if you think we have described something inaccurately anywhere on the site, we would rather hear about it than have it stand. Tell us through the contact form at speechtherapymagic.com/contact, or use the support chat in the corner of any page.

This disclaimer may be updated as the product changes. The date at the top of the page reflects the most recent revision.