01
Capture
Short audio buffers are scored without storing raw speech by default.
Hear every word, tune every lesson.
Prolexia turns oral reading into precise learning signals: live word detection, phoneme guidance, and teacher-ready next steps for each student.
- Response window
- 250 ms
- Detection accuracy
- 85%
- False positives
- -68%
down from 1000 ms
targeted by enhanced matching
through stricter validation
Live reading session
Maya, Grade 1, target /m/
Listening
The Moth and the LampPage 2 of 6
Thebravelittlemothfoundabrightlamp.
Live word-by-word reading
audio windows keep the child in flow
Phoneme-level feedback
sound targets surface before frustration
Adaptive confidence
thresholds shift by word complexity
Reading practice that reacts while the student is still reading.
The essential job is not generating another worksheet. It is hearing a learner's attempt, spotting the sound-level gap, and keeping practice gentle enough to continue.
02
Interpret
Exact, phonetic, and fuzzy matching are weighed against the expected word.
03
Adapt
Stories and practice sets tighten around the sounds that need another pass.
Teacher intervention queue, not another analytics swamp.
Prolexia condenses reading sessions into the next action a teacher can trust: who needs help, which sound is driving the issue, and what to assign next.
Small-group ready
Students cluster by sound pattern, not just by score.
Progress with context
Accuracy, pace, and retry patterns stay connected.
Today's reading signals
3 students need a targeted follow-up
MMaya
confuses /m/ and /n/
Assign moth mini-set
JJon
strong accuracy, slower pace
Advance Fry level
AAri
multi-word drift
Shorten next page
Built for real classrooms, rough audio, and privacy constraints.
The platform already has the hard pieces: local inference options, stricter word matching, phoneme validation, and adaptive processing for lower-power devices.
Hybrid speech engine
Browser recognition, Whisper workers, and fallback paths stay coordinated.
Raw-audio restraint
Session analytics can persist without uploading sensitive recordings.
Mastery model
Progress follows phonemes, Fry level, complexity, and recent confidence.
Teacher-ready output
Each session resolves into an assignment, not just another dashboard chart.
Start with one reader and one sound.
Create an account, assign a short story, and let the next lesson come from the student's actual reading attempt.