
EchoAssist
Speaker-labeled live captions
1 ratings
Free
Rating summary
About
EchoAssist captions in-person conversations live and labels each line with who said it, entirely on your iPhone. Nothing you say is uploaded, because nothing needs to be.
Most live transcription shows what was said. EchoAssist also shows who said it: two speech models run side by side on your device, one recognizing words and one telling voices apart, and each word is attributed to a speaker about half a second after it's spoken.
LIVE CAPTIONS
• Each line labeled by speaker (Speaker 1, Speaker 2), assigned automatically as voices are told apart
• New words appear immediately in a lighter shade and settle in place once the speaker is confirmed, so captions don't jump or reflow
• A haptic tap when the speaker changes, so you can follow turn-taking without watching the screen
• Punctuated and capitalized as it goes
• Works offline once set up
TRANSCRIPTS
• Every session saves automatically as text when you press Stop
• A short on-device summary of each conversation
• Search within a transcript, with every match highlighted and an "N of M" bar to step through them
• Rename recordings and speakers ("Speaker 1" → "Mom")
• Export as plain text through the share sheet
TRANSLATION
Translate a saved transcript into Spanish, French, German, Mandarin, Japanese, or Korean with Apple's on-device Translation. Speaker names carry over, and results are cached so switching languages is instant.
BIGGER SCREENS
• On iPhone Duo's inner display, Past Recordings and Settings show the list and the open item side by side
• In landscape, captions use the full width with controls in a row below
• Whatever you have open stays put as you fold, unfold, or rotate
ACCESSIBILITY
• Every screen follows your system Text Size, including the largest accessibility sizes
• Or set an app-only text size with a slider and live preview
• Light, dark, or system appearance
• A guided walkthrough on first launch, replayable from Settings
PRIVATE BY DESIGN
No account, no servers, no analytics. Speech recognition, speaker detection, summaries, and translation all run on device. Audio is processed in memory and never saved: when you press Stop, only the text remains. EchoAssist is not a voice recorder.
The app itself connects to the internet for one thing: a one-time download of about 800 MB of speech models, from its setup screen before your first session, so pressing Start never triggers a surprise download. You can remove the models in Settings to free the space.
Please caption responsibly. Recording and captioning laws vary by place, and some require everyone's consent. Let people know when a conversation is being captioned.
Summaries require an Apple Intelligence-capable device. On other devices, the summary shows the opening of the transcript instead.
Show more
+1
What's New in EchoAssist
1.2
November 14, 2026


