Dictation Practice
Create a listening exercise from your own sentences and audio clips. Each answer stays off the practice screen until that sentence is submitted. Play or repeat the clip, type what you heard, then review missing and extra text. Every retry adds a separate attempt snapshot. This page does not transcribe speech, certify language accuracy or send audio to a service.
Key features
- User-authored sentence labels, answers and per-sentence local WAV/MP3 clips
- Answer hidden from the practice screen and accessibility tree before submission
- Explicit case, punctuation and whitespace comparison choices
- Repeated playback count and sentence-level missing or extra character segments
- Editable retry with append-only attempt, answer and policy snapshots
- Versioned JSON project and formula-safe CSV attempt report; audio files stay local and must be reattached
How to use
- Load the text example or create a set. Enter a neutral label and the answer sentence for each item.
- Select each item and attach its own WAV or MP3 file. Files are checked locally for extension, type, header, size and decodable duration.
- Choose whether case and punctuation must match and whether whitespace is exact or collapsed. Start practice once every item has audio.
- Listen and type without seeing the answer. Replay as needed, submit, then inspect expected and typed text plus labeled missing and extra runs.
- Choose Retry to edit and resubmit, or Next to continue. Export JSON and CSV from the editor; reattach audio after importing JSON.
Use cases
- Practice writing a teacher-recorded sentence after repeated listening
- Review whether punctuation differences should count in a lesson
- Compare a learner's retries while preserving the answer used at each attempt
- Prepare private listening practice without microphone recognition
Frequently asked questions
Does this record or transcribe my voice?
No. You attach existing local WAV or MP3 audio and type the sentence yourself. The separate Speech to Text page uses microphone recognition; this page compares your typed response with the supplied answer.
Can learners see the answer before submitting?
The practice screen renders only the neutral label, audio controls and response field until submission. Authors can return to the editor to view or change the answer. This is a learning interface, not exam security against browser inspection or an author who knows the answer.
What counts as a correct sentence?
Unicode NFC is always applied. You choose exact or ignored case, exact or ignored Unicode punctuation, and exact whitespace or runs collapsed to one trimmed space. The policy used is saved with every attempt; symbols other than punctuation remain. The comparison is textual, not a grammar or meaning judgment.
What are the colored difference segments?
The page aligns the normalized answer and response by character and labels equal, missing and extra runs in text as well as color. Substitutions appear as adjacent missing and extra runs. Repeated characters can have more than one valid alignment; this display is an editing aid.
Where do audio and attempts go?
Audio stays as a temporary blob URL in this browser tab and is revoked on replacement, removal or exit. The JSON download contains sentences and attempt snapshots but no audio bytes; reattach clips after importing. The CSV includes answers and responses, so keep exports private.
Why will a clip not attach?
Only local WAV or MP3 under 8 MiB with a matching header and a decodable duration of at most three minutes is accepted. A browser may still fail playback for an unsupported codec. The page then reports the playback problem instead of counting a successful repeat.
Privacy
Sentences, audio, answers and attempts stay in this browser tab. No remote recognition, upload or automatic save occurs. JSON and CSV exports contain answers; protect them before sharing. Audio is never embedded in JSON.
Comments & questions