Saywave · Guide
User Guide
Set up dictation, choose how Saywave processes your words, and find the controls you need.
1. Start dictating
Select a supported text field, press ⌃⌥ (Control + Option), speak, then press the shortcut again. Saywave attempts to insert the result at the active cursor. Secure fields and some custom app controls may reject automated insertion.
For push-to-talk, open the menu-bar icon and choose Switch to Push-to-Talk. Saywave then records while you hold ⌃⌥ and stops when you release it.
Change the shortcut in Settings → Shortcut. You can also turn the dictation indicator off there.
Read the dictation indicator
The floating indicator shows the input level, a live text preview, and status messages. Press Esc or click the indicator to cancel. You can drag it to another position, which Saywave remembers.
2. Choose a model and language
Open the menu bar's Model submenu to choose a supported speech engine. The available options include Whisper, NVIDIA Parakeet, and Apple's on-device engine on supported macOS versions. Parakeet v3 is the default. Accuracy and speed vary by language, hardware, audio quality, recording length, and model.
A model may need to download before first use. Saywave can also show Optimizing… while preparing a model for the Mac. Downloaded models can remain loaded while resources allow, which may reduce later switching time.
Use the Language submenu to select your spoken language. Saywave selects a compatible multilingual engine for the language you choose.
3. Choose what happens to the transcript
The Mode submenu controls how Saywave handles the transcript:
- Raw: use the speech model's transcript without cleanup.
- Clean: remove filler words and adjust punctuation and capitalization.
Settings → Cleanup contains the same Raw and Clean choices. Review transcripts before relying on them; speech recognition can make mistakes.
Version 1.3 focuses on transcription. AI rewriting, translation, Voice Edit and new summary or action-item generation are deferred. Previously saved content remains readable and exportable; your saved settings are retained for a future release.
Use spoken formatting commands
Say "new line", "new paragraph", or "bullet point" to insert formatting in supported languages. These commands work in Clean mode and can be disabled in Settings → Cleanup.
4. Use vocabulary and history
Add vocabulary
Settings → Dictionary lets you add names and technical terms or replace a spoken phrase with another value. Whisper models can use terms as recognition hints. Apple Speech and Parakeet apply configured replacements after transcription.
Review local history and statistics
Settings → History stores dictations locally so you can copy them again. Statistics tracks words dictated, speaking time, and words per minute on the device.
5. Capture a meeting
Tell every participant before recording and obtain all consent required by the laws and workplace rules that apply. Do not use Saywave for covert recording. If anyone objects or withdraws consent, do not start the capture. If it has already started, stop it immediately.
Open the menu bar's Meeting submenu and choose Start Meeting Capture. Saywave records microphone and system audio, then creates a speaker-labelled transcript. Choose Stop Meeting Capture when the call ends.
The first capture requires macOS Screen Recording permission because that permission also covers system-audio capture. Saywave does not record or store video. Relaunch the app after granting permission.
Use automatic capture carefully
Auto-Capture in Zoom, Teams & FaceTime starts and stops capture with those apps. It is off by default. Enable it only when your workflow gives every participant advance notice and obtains any required consent before capture begins.
Read, search, and export
Captures appear in Settings → Meetings. Search by app, date, speaker, or transcript text. You can copy the full transcript, copy one voice, or export a Markdown or plain-text file. Click a speaker label to name that voice for the meeting.
Remember voices across meetings is off by default. When enabled, named voice fingerprints stay on the Mac and can be managed or forgotten in the Meetings pane.
Manage saved meetings
New meetings produce transcripts without generated summaries or action items. Older saved summaries remain available. Delete individual meetings from their row or use Clear All.
6. Fix common problems
Dictation does not insert text
Check the menu-bar status. If the microphone captured silence, confirm the input selected in macOS Sound settings or use Saywave's Input submenu. If transcription works but insertion fails, try a standard non-secure text field.
Grant Accessibility keeps appearing
Saywave needs Accessibility permission to insert text. Use the menu bar's Grant Accessibility… item or open System Settings → Privacy & Security → Accessibility, grant access, and try again.
A model is slow the first time
First use can include a download and an optimization step. Duration depends on the model, hardware, available memory, and network connection.