gustavscirulis/teika

Free and open-source speech-to-text notes for iPhone and Apple Watch.

2

stars

5

commits

Swift

primary language

Sep 4, 2026

updated

teika.app

README

Teika — catch the thought before it is gone

Teika is a free and open-source speech-to-text notes app for iPhone and Apple Watch. Tap the circle, speak for as long as you need, then tap again. The transcription is copied to your clipboard and saved in a searchable local library.

How it works

  1. Start from what is closest. Open Teika from the app, Apple Watch, a widget, Control Center, Shortcuts or Siri.
  2. Talk until you are done. Pausing to think does not end the recording.
  3. Paste it now or find it later. Each transcription goes to the clipboard and the on-device notes library.

Privacy

Speech recognition runs on the iPhone. Teika does not upload recordings, transcripts or notes, and it has no account, subscription, ads or in-app analytics. After you approve a one-time model download of about 0.5 GB from Hugging Face, transcription works offline.

After setup, Apple Watch recordings can be queued while the iPhone is unavailable and are transferred through Apple's WatchConnectivity service for local transcription when the connection returns. See the full privacy policy for details.

Screenshots

Record a thoughtReview a captured noteBrowse all notes
Teika recording screen with the record buttonTeika displaying a captured note with a copied-to-clipboard confirmationTeika notes library with saved notes grouped by date

Installation

Download Teika from the App Store, or build it from source.

Requirements

  • An iPhone or iPad running iOS 26.5 or later
  • watchOS 26.0 or later for the companion Apple Watch app
  • Xcode with the iOS 26.5 and watchOS 26 SDKs
  • XcodeGen

Run the app

run.sh generates the Xcode project, lets you choose an available simulator or device, builds Teika and launches it:

./run.sh

You can also pass a destination directly:

./run.sh app "iPhone 17 Pro"

Simulator builds need no signing configuration. For a physical device, provide your Apple Developer Team ID locally:

TEIKA_DEVELOPMENT_TEAM=YOUR_TEAM_ID ./run.sh app "Your iPhone"

The same script can create a signed App Store archive or archive and upload it directly to App Store Connect. Both paths are also available from the picker shown by ./run.sh:

TEIKA_DEVELOPMENT_TEAM=YOUR_TEAM_ID ./run.sh archive
TEIKA_DEVELOPMENT_TEAM=YOUR_TEAM_ID ./run.sh upload

The upload path uses the Apple Developer account signed into Xcode and preserves the version and build number declared in project.yml.

Run the tests

Generate the project, then select a simulator installed on your Mac:

xcodegen generate
xcodebuild test \
  -project Teika.xcodeproj \
  -scheme Teika \
  -destination 'platform=iOS Simulator,name=iPhone 17 Pro'

The live model-download UI test is skipped by default because it downloads about 0.5 GB. Set RUN_MODEL_DOWNLOAD_TEST=YES only when you intend to exercise it.

Run the website

The marketing, support and privacy site is a static Next.js export in site/:

./run.sh site          # development server
./run.sh site build    # production export
./run.sh site preview  # build and serve the export

Acknowledgements

License

Teika is licensed under the GNU General Public License v3.0 or later, with an additional permission for distribution through the Apple App Store. See LICENSE.

Contributors

gustavscirulis/teika

Free and open-source speech-to-text notes for iPhone and Apple Watch.

2

stars

5

commits

Swift

primary language

Sep 4, 2026

updated

teika.app

README

Teika — catch the thought before it is gone

Teika is a free and open-source speech-to-text notes app for iPhone and Apple Watch. Tap the circle, speak for as long as you need, then tap again. The transcription is copied to your clipboard and saved in a searchable local library.

How it works

  1. Start from what is closest. Open Teika from the app, Apple Watch, a widget, Control Center, Shortcuts or Siri.
  2. Talk until you are done. Pausing to think does not end the recording.
  3. Paste it now or find it later. Each transcription goes to the clipboard and the on-device notes library.

Privacy

Speech recognition runs on the iPhone. Teika does not upload recordings, transcripts or notes, and it has no account, subscription, ads or in-app analytics. After you approve a one-time model download of about 0.5 GB from Hugging Face, transcription works offline.

After setup, Apple Watch recordings can be queued while the iPhone is unavailable and are transferred through Apple's WatchConnectivity service for local transcription when the connection returns. See the full privacy policy for details.

Screenshots

Record a thoughtReview a captured noteBrowse all notes
Teika recording screen with the record buttonTeika displaying a captured note with a copied-to-clipboard confirmationTeika notes library with saved notes grouped by date

Installation

Download Teika from the App Store, or build it from source.

Requirements

  • An iPhone or iPad running iOS 26.5 or later
  • watchOS 26.0 or later for the companion Apple Watch app
  • Xcode with the iOS 26.5 and watchOS 26 SDKs
  • XcodeGen

Run the app

run.sh generates the Xcode project, lets you choose an available simulator or device, builds Teika and launches it:

./run.sh

You can also pass a destination directly:

./run.sh app "iPhone 17 Pro"

Simulator builds need no signing configuration. For a physical device, provide your Apple Developer Team ID locally:

TEIKA_DEVELOPMENT_TEAM=YOUR_TEAM_ID ./run.sh app "Your iPhone"

The same script can create a signed App Store archive or archive and upload it directly to App Store Connect. Both paths are also available from the picker shown by ./run.sh:

TEIKA_DEVELOPMENT_TEAM=YOUR_TEAM_ID ./run.sh archive
TEIKA_DEVELOPMENT_TEAM=YOUR_TEAM_ID ./run.sh upload

The upload path uses the Apple Developer account signed into Xcode and preserves the version and build number declared in project.yml.

Run the tests

Generate the project, then select a simulator installed on your Mac:

xcodegen generate
xcodebuild test \
  -project Teika.xcodeproj \
  -scheme Teika \
  -destination 'platform=iOS Simulator,name=iPhone 17 Pro'

The live model-download UI test is skipped by default because it downloads about 0.5 GB. Set RUN_MODEL_DOWNLOAD_TEST=YES only when you intend to exercise it.

Run the website

The marketing, support and privacy site is a static Next.js export in site/:

./run.sh site          # development server
./run.sh site build    # production export
./run.sh site preview  # build and serve the export

Acknowledgements

License

Teika is licensed under the GNU General Public License v3.0 or later, with an additional permission for distribution through the Apple App Store. See LICENSE.

Contributors

Languages

Swift

68.6%

TypeScript

16.9%

Shell

7.1%

CSS

6.2%