Skip to main content
Accessibility MVP · Demo mode

Captions show what was said. ToneCaption shows how it was said.

An accessibility tool for hard-of-hearing users that adds emotion, speaker context, and nonverbal cues to captions — so meaning isn't lost between the lines.

AlexI'm fine.
possibly sarcasticConfidence: 78%
MayaCan we talk?
serious toneConfidence: 91%
[laughter]
JordanThat's great.
excitedConfidence: 88%

The problem with captions today

Live captions tell you the words — but not the sarcasm, the pause before a hard question, the laughter under someone's breath, or whether the room just went quiet. For hard-of-hearing users, that missing layer of meaning is where conversations slip away.

Tone tags

Short, plain-language labels like [possibly sarcastic], [serious tone], or [excited] appear beside each caption.

Nonverbal context

Captures [laughter], [long pause], [multiple people talking], so you don't miss what happens between words.

Personal alerts

Get flagged when your name is spoken or when a question is directed at you.

Speaker labels

Know who's talking, even when several people are on the call.

Confidence indicators

Every tone tag is an estimate — see the model's confidence and decide for yourself.

Accessible by default

Large, high-contrast type. Keyboard navigable. Adjustable font size and contrast.

Who it helps

  • · Hard-of-hearing and Deaf users in video calls
  • · Students following recorded lectures
  • · Anyone in a noisy environment
  • · Non-native speakers picking up on tone

Privacy first

Your audio should be processed securely. A production version will let you choose local or secure cloud processing. Demo mode does not store audio.

Note: ToneCaption is not a medical or diagnostic tool. Tone labels are AI-generated estimates and may not always be correct.