Fast, High-Quality Cloud Setup
Cloud providers offer the fastest processing and highest accuracy for Stenox dictation. This guide shows you the recommended configurations for optimal results.Why Use Cloud Providers?
Ultra-fast processing
Sub-second transcription and enhancement (5-10x faster than local).
Higher accuracy
State-of-the-art models optimized for production use.
No local storage
No need to download multi-GB models.
Works on any Mac
Intel and Apple Silicon, old and new.
Recommended Cloud Configurations
Option 1: Stenox Cloud (Most Recommended)
Best for: Fastest setup, best out-of-the-box experience Configuration:- Transcription: Stenox Cloud Transcription
- Enhancement: Stenox Cloud AI Enhancement
1
Get Plus or Lifetime Plus
Visit stenox.app/pricing and choose your plan
2
Activate License
Settings → License → Sign in with your purchase email
3
Select Stenox Cloud
Settings → Transcription → Provider: Stenox Cloud
Settings → AI Enhancement → Provider: Stenox Cloud
4
Start Dictating
That’s it — no API keys, no configuration, just works
- Fine-tuned for Stenox — optimized specifically for voice dictation workflows
- Zero configuration — no API keys to manage, no accounts to create
- Consistent quality — we maintain and update the models for best results
- Fast and accurate — production-grade speed and accuracy
Learn More
Full Stenox Cloud details
Option 2: Maximum Free Credits (BYOK)
Best for: Getting started free, maximum free usage Configuration:- Transcription: DeepGram Nova-3 ($200 free credits)
- Enhancement: Google Gemini 2.5 Flash (1,500/day free)
1
Get DeepGram API key
- Visit console.deepgram.com
- Sign up (no credit card required)
- Create API key
- Get $200 in free credits
2
Configure DeepGram in Stenox
- Settings → Transcription tab
- Provider: DeepGram
- Paste API key
- Model: Nova-3 (best accuracy)
3
Get Google Gemini API key
- Visit aistudio.google.com
- Sign in with Google (no credit card)
- Create API key
- Get 1,500 free requests/day
4
Configure Gemini in Stenox
- Settings → AI Enhancement tab
- Provider: Google Gemini
- Paste API key
- Model: Gemini 2.5 Flash (fastest)
Option 3: Groq Stack (Single API Key)
Best for: Simplicity, one account for everything Configuration:- Transcription: Groq Whisper v3-turbo
- Enhancement: Groq Llama 3.3 70B or DeepSeek R1
1
Get Groq API key
- Visit console.groq.com
- Sign up (no credit card required)
- Create API key
- One key for both transcription and enhancement!
2
Configure Groq Transcription
- Settings → Transcription tab
- Provider: Groq
- Paste API key
- Model: whisper-large-v3-turbo
3
Configure Groq Enhancement
- Settings → AI Enhancement tab
- Provider: Groq
- Paste same API key
- Model: Llama 3.3 70B Versatile
- Single API key management
- Ultra-fast LPU inference
- Ongoing free tier (doesn’t expire)
- Simple setup
Option 4: Best Quality (Paid BYOK)
Best for: Professional use, maximum accuracy Configuration:- Transcription: DeepGram Nova-3 or AssemblyAI Universal-1
- Enhancement: Google Gemini 2.5 Pro
- Professional transcription services
- Critical accuracy requirements
- Business or enterprise use
- After free credits are exhausted
Performance Comparison
| Configuration | Speed | Accuracy | Setup | Cost | Best For |
|---|---|---|---|---|---|
| Stenox Cloud | ⚡⚡⚡ | ⭐⭐⭐⭐⭐ | Instant | Plus/Lifetime Plus | Best experience, zero config |
| DeepGram + Gemini | ⚡⚡⚡ | ⭐⭐⭐⭐ | 10 min | $$ after free | Maximum free credits |
| Groq Stack | ⚡⚡⚡ | ⭐⭐⭐⭐ | 5 min | Free tier | One API key simplicity |
| AssemblyAI + Gemini | ⚡⚡⚡ | ⭐⭐⭐⭐ | 10 min | $ | Accent optimization |
| Local (WhisperKit + MLX) | ⚡ | ⭐⭐⭐ | 5 min | Free forever | Privacy, offline |
Detailed Provider Configurations
DeepGram Nova-3 Settings
Recommended configuration:| Setting | Value | Why |
|---|---|---|
| Model | Nova-3 | Latest, best accuracy |
| Language | English (or auto-detect) | Optimize for your language |
| Streaming | Enabled (automatic) | Fastest results |
| Punctuation | Enabled (automatic) | Auto-punctuation |
- Provider: DeepGram
- Model: Nova-3 or Nova-3-English
- Language: English or Auto-detect
Google Gemini 2.5 Flash Settings
Recommended configuration:| Setting | Value | Why |
|---|---|---|
| Model | Gemini 2.5 Flash | Fastest with excellent quality |
| Custom prompt | Default or Professional | Based on use case |
| Temperature | Default | Stenox handles this |
- Provider: Google Gemini
- Model: Gemini 2.5 Flash
- Custom prompt: (optional, see below)
Enhancement Prompts
For BYOK providers, you can configure custom enhancement behaviors:- Default (General Use)
- Professional Email
- Technical Documentation
Multi-Profile Cloud Setup
For maximum flexibility, create profiles for different scenarios:Profile Strategy
1
Work Email Profile
- Name: “Work Email”
- Transcription: DeepGram Nova-3
- Enhancement: Gemini 2.5 Flash
- Custom prompt: Professional email tone
- Auto-activate: When Gmail or Outlook is active
2
Quick Notes Profile
- Name: “Quick Notes”
- Transcription: Groq Whisper (fastest)
- Enhancement: None (skip for speed)
- Auto-activate: When Notes app is active
3
Documentation Profile
- Name: “Documentation”
- Transcription: DeepGram Nova-3
- Enhancement: Gemini 2.5 Flash
- Custom prompt: Technical documentation
- Auto-activate: When VSCode or similar is active
4
Fallback Local Profile
- Name: “Offline”
- Transcription: WhisperKit
- Enhancement: MLX or None
- Manual activation: When internet unavailable
Usage Optimization
Maximize Free Credits
DeepGram ($200 credits):- Use for important dictation (work emails, documents)
- Switch to Groq or local for casual notes
- Track usage in DeepGram console
- More than enough for most users
- If you hit limit, create “no enhancement” profile
- Resets daily at midnight PST
- Use as primary option — no API costs
- Rate limits refresh quickly
- Good for privacy-conscious users who prefer not to create accounts
Cost Control
After free tiers, control costs:Use Groq free tier as primary
Use Groq free tier as primary
Groq’s ongoing free tier is sufficient for most personal use. Only upgrade if you consistently hit rate limits.
Skip enhancement for quick notes
Skip enhancement for quick notes
Create a profile without AI enhancement for quick personal notes. Saves on Gemini API calls.
Switch to local for high-volume
Switch to local for high-volume
If you dictate heavily (2+ hours/day), switch to WhisperKit + MLX to avoid costs.
Use AssemblyAI after DeepGram
Use AssemblyAI after DeepGram
AssemblyAI costs only $0.0009/min after free credits - very affordable for continued use.
Troubleshooting Cloud Setup
Slow processing despite cloud providers
Slow processing despite cloud providers
- Check your internet connection speed
- Try switching to different model (e.g., Flux instead of Nova-3)
- Check provider status pages for outages
- Test with local provider to isolate issue
Poor transcription accuracy
Poor transcription accuracy
- Ensure you’re using Nova-3 or latest models
- Check microphone input device (Settings → Audio tab)
- Reduce background noise
- Speak more clearly or use better microphone
- Try AssemblyAI if you have strong accent
API errors or authentication failures
API errors or authentication failures
- Verify API keys are correct (no extra spaces)
- Check API key status in provider dashboards
- Ensure you have remaining credits/quota
- Try creating new API keys
Hit rate limits frequently
Hit rate limits frequently
- Create local fallback profile (WhisperKit + MLX)
- Set up multiple cloud providers and rotate
- Consider upgrading to paid tier if needed
Internet Connection Requirements
Minimum requirements:- Speed: 1 Mbps upload, 1 Mbps download
- Latency: < 100ms preferred
- Stability: Consistent connection (not intermittent)
- Speed: 5+ Mbps upload/download
- Latency: < 50ms
- Type: Wi-Fi or Ethernet (not cellular for best results)
Cloud processing sends ~1 MB per minute of audio. Most home internet connections are more than sufficient.

