Audio

An Audio question asks the candidate to answer by speaking. The candidate records with their microphone directly in the browser — no software to install — previews the recording, and submits it. Audio answers are the backbone of language speaking tests, pronunciation checks, and verbal reasoning assessments.

Recording Controls

• Time limit — cap each recording's length (default 60 seconds) or allow unlimited recording time. When limited, recording stops automatically at the limit.
• Attempt limit — how many times the candidate may re-record (default 1 attempt) or unlimited retries. A single attempt makes the response spontaneous — essential for speaking exams; multiple attempts suit lower-stakes practice.
• Microphone selection — candidates with several microphones pick the right one from a selector; input level is metered live so they can confirm the mic works before recording.

How Recordings Are Handled

Recording works in every modern browser, desktop and mobile — including Safari and iPhone/iPad. The browser records in its native audio format and the file uploads immediately when the candidate stops; the candidate can play it back before moving on.

Evaluation

• AI Evaluation — the AI first transcribes the spoken content, then evaluates the transcription against your grading prompt. Transcription and feedback follow the language set on the question (Details tab), so a French speaking question is transcribed and evaluated in French. See AI Evaluation.
• Rubric scoring — attach a rubric (fluency, pronunciation, content, grammar) for consistent human grading. See Rubric Scoring.
• Manual review — reviewers listen to the recording in the evaluation screens and score directly.

Common Features

Audio questions support all shared question features — sidebar layout (show the reading passage or image being described), hints, scoring multipliers and dimensions, folders and tags. See Common Question Features.

For listening-then-speaking tasks, put the audio clip in the question content with a play limit (see Content Editor → Audio Clips) and set the answer's attempt limit to 1 — the candidate hears the stimulus a controlled number of times and answers spontaneously.

Tip