For Speech-Language Pathologists

Clinical collaboration around intelligibility and participation

Helping you be understood, not just heard

DysVoxa Neurospeech Platform is designed as a practical support layer for real-world communication. SLP involvement is only for app testing, not for user profile enrollment or outcome tracking.

Enrollment workflow

Guided voice sample capture to establish personalized pronunciation profile.
Domain phrase lists protect medication and therapy terminology from incorrect rewrites.
5-minute baseline setup, then iterative improvements with clinical feedback.

Utterance strategy

Target phrase length: 5-10 words, usually 3-7 seconds.
Silence threshold around 700ms, with dynamic adaptation when enabled.
Tiered correction balances intelligibility gains and identity preservation.

Sample voice profile and voice card

These examples are taken from current project templates and are used during enrollment and correction prompt calibration.

Sample profile

Speaker ID: speaker_sca_20260318
Dysarthria Type: ataxic
Dysarthria Severity: moderate
Disease/Etiology: spinocerebellar ataxia

Open full sample profile

Voice card template

Speaker ID: [SPEAKER_ID]
Dysarthria Severity: [mild|moderate|severe]
Top confusion patterns: r->l, t->d, s->z
Review schedule: every 3-6 months

Open full voice card template

Supported dysarthria phenotypes

Ataxic

Flaccid

Hyperkinetic

Hypokinetic

Mixed

Spastic

Pilot protocol framing

Initial cohort: 5 participants, diverse etiologies and severity profiles.
Measure baseline and follow-up intelligibility using Word Error Rate (WER) and listener-based understanding ratings.
Assess psychosocial outcomes: speaking confidence, participation, and communication burden.

Compliance direction

Intended path

EU Medical Device Regulation (MDR) Class I positioning (non-diagnostic communication aid)

General Data Protection Regulation (GDPR) Article 9-aware voice data handling model

Technology Performance Evaluation Card Template

Printable card for technology performance assessment before, during, and after processing. Full Card (Print/PDF)

Patient & Intervention

Pre-date:: ________
Post-date:: ________
ID/Age/Gender:: ________
Diagnosis/Severity:: ________
Usage duration:: ___ weeks

Subjective

Pre VHI-10:: /40 ___
Post VHI-10:: /40 ___

Before / After Comparison

Metric	Before	After
ASR Word Error Rate (%)	___	___
Clarity (1-5)	___	___
Self-Assessment (1-5)	-	___
Intelligibility (1-5)	-	___

Comments:

Evaluator: ________ Date: ________

Neuroplasticity, recovery windows, and real-time support

After stroke, brain injury, or progressive neurological disease, neuroplasticity offers a window of recovery. DysVoxa supports this window by reducing communication barriers during rehabilitation:

Early recovery: In the first weeks and months post-stroke or TBI, intensive practice (including speech practice) correlates with better neurological outcomes. Reducing communication frustration removes a barrier to participation.
Confidence and engagement: When patients can be understood in real-time, they're more likely to engage in social and professional activities, which reinforces neuroplastic recovery pathways.
Progressive disease: For conditions like ALS or Parkinson's, DysVoxa offers a non-pharmacological intervention to slow perceived decline in intelligibility and preserve participation as motor control changes.
Outcome measurement: Combined with your traditional therapy metrics (articulation scoring, voice quality, duration), DysVoxa output (WER, real-world intelligibility ratings) offers a second stream of evidence that your interventions are working.

Clinical framing: Position DysVoxa as a supportive tool that doesn't replace speech therapy—it extends and amplifies it by enabling real-time practice and participation in daily communication contexts where formal articulation drills would not occur.

Current quality-pass state

Current reference set includes 120 labeled utterances and 480 reference tokens.
Quality harness runs 3 comparable modes: no_correction, correction_no_bias, correction_with_bias.
Outputs are tracked in machine-readable summaries for internal comparison.
Iteration remains active: correction logic and dictionaries are still being refined between passes.

SLP pilot inquiry

Use this form to discuss pilot collaboration, evaluation metrics, or enrollment workflow fit.