Voice notes with transcripts, built into the chat

The problem with plain voice messages

Voice messages are warm and quick to record, and they are slow to receive. You cannot skim one, search for the part about Friday, or glance at it in a quiet room. Many people respond by asking the sender to type instead, which defeats the point.

A transcript removes most of that cost. The audio is still there for tone and for the moments when listening is easy, and the text is there for everything else.

How it is planned to work

You hold the microphone button in the message box, speak, and let go. The note is sent straight away. A short time later its transcript appears beside it.

• The microphone sits next to the text field, so recording is one gesture.
• The transcript is attached to the message, not to a separate screen.
• A waveform shows how long the note is and where the pauses are.
• Playback progress is drawn on the waveform, so you can see where you are.

Search and quoting

Because the transcript is text, it behaves like any other message. It can be searched, and a line from it can be quoted in a reply. A decision made out loud in a note is no longer lost in a recording nobody wants to scrub through.

Transcripts will not be perfect, particularly with background noise or unusual names. The audio stays available so the original is always one press away.

Respecting the people who record

A voice note is personal. Yappio's design keeps notes inside the channel or conversation they were sent to, and the details of how transcription is processed and stored will be written down on a public page before launch, in plain language.

What a transcript is for

The transcript is not a replacement for the recording. It is there so you can decide what to do with a note before you press play: read it in a meeting, skim it on a train, or skip it because it only says that Friday is fine. When the tone matters, the audio is still one tap away and still the original.

A transcript also makes a note findable later. Without one, a voice message is a sealed box that nobody can search. With one, the line about the folding chairs can be found by typing the word chairs.

Limits worth knowing about

Automatic transcription makes mistakes. Names, accents, background noise and technical words are all hard for any system, and we expect Yappio's transcripts to be wrong sometimes. The design therefore shows the transcript as a convenience, never as the official record of what someone said, and the recording stays attached so people can check.

We are also weighing where transcription should run and how long audio and text are kept. Those choices affect privacy and cost, and they are not decided. The privacy notice describes only what the website does today. Before the app opens, the policy for voice data will be published and you will be able to read it before you record anything.

Where it helps most

Communities spread across time zones, where a spoken update can be read the next morning. People who are more comfortable talking than typing. Study groups that want to capture a quick explanation. Friends who send a lot of voice notes and want to be able to find them again.