Skip to main content
SayfieldPrivate betaSign in

Say it out loud. See what you actually said.

Sayfield is a speaking-practice app for anyone who speaks English, as a first language or not. Answer a question out loud for about a minute, then play back your recording with every word written out.

Already invited?

A real meeting, written out word for word

Um, just so we can work off of them, see how fashions have changed. Um, so I'll list the three most important aspects that I've come across, um, and they, they're each more important than the one that comes after it.
FillerPauseTap a word to hear its phrase.

12 seconds of a real meeting, written out by Sayfield itself: fillers (um, uh) outlined, pauses marked where they fell.

How it works

  1. Pick a question you'd argue about

    Answer out loud, for about a minute. The questions are written to get a reaction: Man or bear? Sell me this pen. Could you land the plane?

    The question deckNot in the app yet
  2. Read back what you said, ums and pauses included

    The transcript shows the ums and the pauses along with the words, each at its place in the recording.

    The close-up: words on the waveform, pauses and fillers markedNot in the app yet
  3. Try one thing, then answer again

    Short notes on your answer will quote your own words back to you: one line worth keeping, and one thing to try. Then you answer the same question again (Take 2) and hear both.

    A thing to try, then Take 1 and Take 2

What happens to a recording, step by step

Where Sayfield is going

Everything you can hear in speech, written down.

Four things you can hear in speech, drawn one by one on the meeting clip you can play at the top of the page.

Words. Each word at its place in the recording.Fillers. Um and uh, and the same sounds in other languages: euh, äh, ııı.Pauses. Silences, measured from the sound itself.Pitch. How the voice rises and falls as it goes.

  1. WordsEach word at its place in the recording.
  2. FillersUm and uh, and the same sounds in other languages: euh, äh, ııı.
  3. PausesSilences, measured from the sound itself.
  4. PitchHow the voice rises and falls as it goes.

Today the app shows words, fillers and pauses. Pitch isn't shown yet.

For developers and researchers

The same layers, as data.

Sayfield is building its own speech models. Each will take audio and return one kind of mark (a label on the transcript, like a pause or where the voice falls), all on one shared timeline. None is released yet.

  1. Words as said

    In
    The 12-second meeting clip
    Out
    across, um,filler · 8.2 s and they, they'rerestart (marked by hand: no model finds these yet)
    Why it matters
    Most speech-to-text tools delete the ums. Interview research and call coaching need them.
  2. Timing and pauses

    In
    The 12-second meeting clip
    Out
    changed.2.8 spause 0.9 sUm,4.0 spause 0.4 sso5.1 s
    Why it matters
    How long each pause lasts and how fast the words come, for speech research and subtitle timing.
  3. Phrasing and endings

    In
    The 12-second meeting clip
    Out
    fashions have changed.voice fallsthat I've come across,voice risesThese endings are Sayfield's guess from simple rules, not from a trained model.
    Why it matters
    Where a phrase ends and how the voice moves there, for voice assistants that need to know when someone has finished.

Built on a review of the research: how everyday speech is written down word for word, how pauses are marked, and which feedback on speaking helps. The research behind it

Private beta

Sayfield is invite-only for now. The first invitations go to a small group.

Invited? Open the invite link you were sent to come in.

To ask about Sayfield, write to hello@sayfield.app.

Write to Sayfield

Want an invite, working on speech, or writing about it? Send a note. The founder reads every message.

Only used to reply to you.

What is it about?

Your message and email stay on Sayfield's server and only the founder reads them. Write again from this form any time to have them deleted.