Transcription Quality & Performance

Several factors affect how fast and accurate StarWhisper transcribes your speech. This guide covers the most common issues and how to resolve them.

AI Text Formatting Adds Delay

If you have AI Text Formatting enabled (a Pro feature), every transcription is sent to an AI model for post-processing after Whisper finishes. This adds 1–3 seconds of extra delay on top of the normal transcription time.

What AI Text Formatting Does

• Adds paragraph breaks where natural topic changes occur
• Fixes run-on sentences and minor grammar issues
• Can format text as emails, notes, or letters (depending on your chosen style)

When to Disable It

Consider turning off AI Text Formatting if:

• Speed is more important than polish (e.g., quick notes, chat messages)
• You notice a noticeable pause after recording stops before text appears
• You prefer to format text yourself

Known Limitation

Because AI Text Formatting sends your transcription through an AI language model, it may occasionally alter meaning or rephrase sentences in unexpected ways. If you notice the output doesn't match what you said, try disabling this feature.

Wrong Microphone Selected

If StarWhisper is using the wrong microphone, transcription quality can suffer dramatically. Common symptoms include:

Symptom	Likely Cause
Many missing or incorrect words	Using a distant or low-quality microphone
Transcription picks up background noise	Using laptop built-in mic instead of headset
No transcription at all (silent recordings)	Selected mic is disconnected or muted
Microphone changes after reboot	USB devices get new IDs after restart

How to Fix

1. Right-click the StarWhisper circle → Settings → Audio tab
2. Check which microphone is currently selected in the dropdown
3. Select the correct microphone (your headset, USB mic, etc.)
4. Click Test Microphone to verify it works

Tips for Best Microphone Quality

• Use a dedicated microphone — headset or USB mics outperform laptop built-in mics
• Position correctly — 6–12 inches from your mouth
• Minimize background noise — close windows, turn off fans during recording
• Avoid Bluetooth for dictation — Bluetooth audio codecs compress audio and reduce quality

GPU Load Affects Transcription

When GPU acceleration is enabled, StarWhisper uses your graphics card to run the Whisper AI model. This is significantly faster than CPU-only transcription, but if your GPU is already under heavy load, it can cause problems.

Symptoms of GPU Overload

• Transcription takes much longer than usual
• Words are garbled, repeated, or missing
• StarWhisper becomes unresponsive during transcription
• You see errors mentioning "CUDA" or "out of memory"

Common GPU-Heavy Activities

Activity	GPU Impact
Gaming	Very high — typically uses 90–100% GPU
Video editing / rendering	High — GPU encoding uses significant resources
AI image generation (Stable Diffusion, etc.)	Very high — saturates GPU memory and compute
Multiple browser tabs with video	Moderate — hardware video decode uses some GPU
Normal desktop use	Low — no impact on StarWhisper

How to Fix

1. Close GPU-heavy applications before dictating (games, video editors, AI tools)
2. Check GPU usage: Press Ctrl + Shift + Esc → Performance tab → GPU. If usage is above 80%, other apps are competing with StarWhisper.
3. Use a smaller model: Switch from Medium or Large to Small or Tiny in Settings → Model. Smaller models need less GPU memory.
4. Disable GPU acceleration temporarily: Settings → Advanced → turn off "GPU Acceleration". This falls back to CPU, which is slower but doesn't compete with other GPU tasks.

Choosing the Right Model

The Whisper model you choose directly affects both speed and accuracy:

Model	Size	Speed	Accuracy	Best For
Tiny	75 MB	Fastest	Basic	Quick notes, simple commands
Base	142 MB	Fast	Good	Everyday dictation
Small	466 MB	Moderate	Very good	Professional dictation (recommended)
Medium	1.5 GB	Slower	Excellent	Technical/medical content, non-English

If you're experiencing slow transcription, try a smaller model. If accuracy is more important than speed, try a larger one.

Quick Reference

Problem	Cause	Fix
Long delay after recording	AI Text Formatting enabled	Disable in Settings → Text
Poor accuracy / wrong words	Wrong microphone selected	Check Settings → Audio
Slow transcription	GPU busy with other apps	Close games / video editors
CUDA / GPU errors	Not enough GPU memory	Use smaller model or disable GPU
Good speed but low accuracy	Model too small	Upgrade from Tiny to Small
Mic changes after reboot	USB device ID reassignment	Re-select mic in Settings