User Manual
Master the art of offline voice utility. (v3.3.2)
Voice to Text (Listen)
The core power of OneSentence is converting your spoken thoughts into text instantly, without ever needing an internet connection.
Trigger
Press ⌘ + 5 to start recording. You will see the menu bar icon change to show an exclamation badge.
Speak
Speak clearly into your microphone. Since OneSentence uses a local implementation of OpenAI's Whisper model, you can speak in any major language with total privacy. The default recording limit is 300 seconds, but this can be adjusted in Settings.
Stop
Press ⌘ + 5 again to stop. The app will immediately begin offline processing.
The "Success" Sound
Wait for a subtle completion tone. This sound is your confirmation that the AI has finished transcribing and your text is ready.
Paste Anywhere
The transcribed text is now on your system clipboard. Just press ⌘ + V in any app to paste your thoughts.
Text to Speech (Read)
Transform any text on your screen into a professional narration using pro-audio AI voices.
How to trigger:
- Highlight any text in any application (Browser, PDF, Email, etc).
- Press ⌘ + 6.
- OneSentence will automatically copy the selection and begin speaking using the Supertonic engine.
- To stop early, just press ⌘ + 6 again.
Power Features
Smart Punctuation
OneSentence automatically polishes your transcribed text. It adds missing periods to headers, fixes encoding errors, and caps repetitive marks (like !!) to ensure the text remains clean and readable for both you and the voice engine.
Adjustable Timeout
By default, recording stops automatically after 300 seconds to prevent accidental runaway captures. You can increase or decrease this in the General settings to match your dictation style.
Repeat transcription back to me
When enabled in Settings, OneSentence will automatically read your transcribed text back to you immediately after transcription finishes. This is perfect for eyes-free confirmation of your dictation.
Settings Reference
General
Launch at Login
Keep OneSentence ready in your menu bar every time you start your Mac.
Global Hotkeys
Record custom key combinations for the "Listen" and "Read Selection" commands.
Playback Speed
Adjust the narration speed from 0.5x to 2.0x to suit your listening preference.
Diffusion Steps
Control the quality of AI voice synthesis. Higher steps (up to 30) result in more natural speech but require more processing time.
Smart Punctuation
Collapses excessive punctuation (like changing "!!!" to "!!") and automatically adds periods to headers and list items during transcription. This ensures the AI voice engine takes natural pauses where intended.
Fix Mojibake
Detects and corrects text encoding errors—specifically fixing cases where symbols like curly quotes or apostrophes are replaced by strange character sequences (e.g., converting ÒQuoteÓ back to “Quote”).
Repeat transcription back to me
Automatically reads the transcribed text aloud immediately after a \"Listen\" command completes. The standard completion sound is suppressed when this is enabled.
Recording Timeout
Configure how long the app should listen before automatically stopping. Adjustable from 10 seconds to 10 minutes.
Enable Sound Feedback
Toggles whether subtle tones are played when starting recording, timing out, or encountering errors. Enabled by default.
Voices
OneSentence comes pre-installed with 10 professional AI voices (5 Male, 5 Female). All voices are available immediately for offline narration without additional downloads.
-
Instant Selection: Choose any voice from the curated list. Your selection is applied instantly across the entire system.
-
Zero Latency: Because all voice parameters are already on your Mac, there is no waiting time when switching between different personas.
Models
Transcription Language
OneSentence defaults to English, but
supports dozens of languages. This setting tells the Whisper engine exactly which
language you are speaking so it can produce the correct character set (e.g.,
Cyrillic, Kanji, or Latin script).
You can choose "Auto" to have the model discern the language
automatically from your first few words, but selecting a specific language improves
transcription speed and prevents detection errors.
Whisper Model Selection
Choose the engine that powers your transcription. The Large-v2 model offers industry-leading accuracy, while smaller models provide near-instant results on older hardware.
On-Demand Downloading
To keep the initial app size small, Whisper transcription models are downloaded securely only when you first select them. Once a model is on your device, it remains available for permanent offline use.
Audio
Precision control over your hardware. Switch between any connected audio device without restarting the app.
Maintenance
Built-in tools to manage your local data and application lifecycle. Both uninstallation options include an optional feedback survey to help us improve the product.
Reset Settings
Restores all application preferences and hotkeys to factory defaults. This is useful if you want a fresh configuration but want to keep your downloaded AI models.
Uninstall
Safely deletes the 2-4GB of AI model data from your disk. Your settings are preserved, making this ideal if you need to free up space temporarily.
Reset & Uninstall
The most comprehensive cleanup option. This wipes all models, deletes all settings, and automatically moves the OneSentence application bundle to the Trash.
Enable Debug Logging
When enabled, detailed logs and recorded audio files are written to the application data folder for troubleshooting. You can open the log folder directly from this section.