Privacy-first
Local transcription is the default path - nothing is uploaded unless you say so.
Loading Sawoice
Privacy-first speech to text
Speech to text that
stays on your device.
Sawoice transcribes across 98+ languages with local AI first - and lets you switch on Cloud Boost only when a workflow needs the extra capacity.
Local by defaultCloud Boost is opt-in98+ languages
Workspace Ready Parakeet-TDT
SawoiceRecord
Capture speech and turn it into text with privacy-first AI.
Listening
Stop ⌘R Copy Send to App
Latest transcript
Today, 10:42 AM · 33sI want to redesign the whole UI for the application - a modern, premium look with a dark theme and amber accents. Local transcription stays on this device, and Cloud Boost only runs when I explicitly turn it on for a heavier workflow.
Today's insights
618
Transcriptions
24,017
Words
116
Avg WPM
Quick actions
Open audio fileImport from clipboardAdd to dictionary
Blazing fast
Real-time transcription with on-device latency.
Local-first processing
Runs fully offline, with Cloud Boost available when you choose it.
Custom dictionary
Add your terms and phrases for personalized accuracy.
98+ language coverage
Dictate naturally across global languages and accents.
How it works
From voice to text in three simple steps.
Sawoice is designed to be beautifully simple so you can focus on what matters.
1
Record
Capture speech with crystal-clear audio using Sawoice.
Press ⌘R anywhere
2
Transcribe
Local AI models process audio first, with Cloud Boost available when enabled.
On-device by default
3
Export & Use
Copy, save, or send your transcript where you need it.
Any app, any format
Language coverage
Dictation that travels across 98+ languages.
Capture everyday speech, meetings, and long-form dictation across a broad global language set without leaving the local-first workflow.
EnglishHindiSpanishFrenchGermanJapaneseArabicTamilPortugueseKoreanItalianMarathiUrduTeluguUkrainianMalay
MandarinRussianDutchTurkishVietnamesePolishThaiBengaliSwedishGreekHebrewIndonesianCzechRomanianFinnishDanish
Local-first
Keep everyday dictation on your machine by default.
Speech ready
Built for real voice notes, meetings, and long-form dictation.
Global reach
Market, write, and work across teams without switching tools.
System requirements
What you need to run Sawoice.
Sawoice runs local AI models on-device. Requirements vary by model - start with Lite for the lowest footprint, or use Swift for the broadest language coverage.
macOS
Minimum
- OS:macOS 13 Ventura
- Chip:Apple M1 or later
- Memory:8 GB unified memory
- Storage:2 GB + model size
Recommended
- OS:macOS 14 Sonoma or later
- Chip:Apple M2 or later
- Memory:16 GB unified memory
- Storage:8 GB free
Coming soon
Windows
Minimum
- OS:Windows 10 (22H2)
- CPU:x64 with AVX2 support
- Memory:8 GB RAM
- Storage:2 GB + model size
Recommended
- OS:Windows 11
- CPU:Intel 12th Gen+ or AMD Ryzen 5000+
- Memory:16 GB RAM
- Storage:8 GB free
Per-model memory requirements
Total system memory needed with the model loaded and recording active.
| Model | macOS RAM8 GB variant | Windows GPU RAMDedicated graphics card |
|---|---|---|
Lite Smallest local English model | 8 GB | 2 GB GPU |
Core Balanced local English model | 8 GB | 2 GB GPU |
Pro Higher-accuracy local English model | 8 GB | 3 GB GPU |
Swift Fast local multilingual model | 8 GB | 3 GB GPU |
Simple, transparent pricing
Choose the plan that fits your workflow.
Start with private local dictation across 98+ languages and add Cloud Boost when your workflow needs it.
Lifetime
Lifetime local-only access.
$100one-time
- All local models, forever
- Private on-device processing
- One-time payment, lifetime access
Pro
Local privacy plus Cloud Boost for daily work.
$50$25/ month
- All local models included
- Cloud Boost transcription and Read Aloud
- 600,000 Cloud Boost words per month
Max
Fair-use Cloud Boost for heavy daily dictation.
$99$69/ month
- Everything in Pro
- Sawoice Live streaming transcription
- 4,000,000 Cloud Boost words per month under fair use