For people who use English at work

Your conversations become your next useful practice.

Record yourself or upload your own audio. Confirm what you said, choose a topic to practise, and answer a new question out loud.

Early pilot: automatic error feedback is still being validated. You can choose a practice topic yourself.

Your own voice only. Up to 30 minutes or 100 MB per file.
Demo · fictional speaker
1 · What you said · 00:41.6

“Yesterday I didn't went to the meeting.”

One sentence, five moments

From something you said to something you can say better.

  1. 1
    You bring your own speech.

    Record on your computer or upload your own file. Choose up to five ranges where you are speaking. Coverage is shown on every result.

    00:0023:40
    2 ranges · 4:20 of 10:00 allowed · covers 18% of the recording
  2. 2
    You confirm the words first.

    If the transcript misheard you, you fix the words, and anything built on them is withdrawn.

    You said · 00:41.6

    “Yesterday I didn't went to the meeting.”

    1Is this exactly what you said?
  3. 3
    Then you choose what to practise.

    Choose a topic to build your lesson. Automatic corrections are still being validated.

    Yesterday I didn't go to the meeting.

    Example · 00:41.6Base form after didn't
  4. 4
    A short lesson, then a new answer out loud.

    A rule, a reading, a few tasks, and a new 30 to 60 second answer. Spoken assessment is still being validated.

    By Tuesday afternoon, the plan had changed. The payment test didn't pass, and the designer didn't get the final copy in time.

    Speak: explain what didn't happen, and what you did instead.
  5. 5
    Tomorrow a fresh prompt. Later, real evidence.

    Practice results and your real conversations are kept apart. Fewer mistakes or a missing phrase is never counted as progress.

    Practicedidn't + base form used correctly 2 times
    Real conversationNot enough data yet
Designed for your own voice

No bot in your calls. No live advice while you talk. No scores, levels or streaks.

A microphone can pick up people nearby or sound from speakers. Use headphones and select only your own speech.
Scheduled deletion
24 hSelected source audio, or sooner after review.
24 hFailed or abandoned uploads.
30 daysShort evidence clips.
YouConfirmed lesson text, until you delete it.

Outages can delay cleanup.

How retention works
talklooma.comEvery example on this page is fictional demo data.