#Model Download Errors

Troubleshoot issues with downloading, loading, and running Whisper and LLM models.

#Download Fails

Model download stops or fails partway through.

#Check Internet Connection

1. Verify you're connected to WiFi or ethernet
2. Test internet by opening a website
3. If no internet, reconnect and retry download

Note: Model downloads can take 30+ minutes depending on file size and connection speed. Don't sleep computer during download.

#Check Disk Space

Downloads require free space equal to the model size:

Check available space:

1. Go to Settings → Storage
2. Look for "Available space"
3. If less than model size, delete old meetings or files

Free up space:

• Delete old meetings: Go to Meetings → Select meetings → Delete
• Empty Trash: Right-click Trash / Recycle Bin and select Empty (macOS) or Empty Recycle Bin (Windows)
• Move files to external drive

#Verify Download Path

Clearminutes saves models to:

• macOS: ~/Library/Application Support/Clearminutes/models/
• Windows: %APPDATA%\Clearminutes\models\

Ensure this folder exists and you have write permission:

macOS:

mkdir -p ~/Library/Application\ Support/Clearminutes/models/
ls -la ~/Library/Application\ Support/Clearminutes/models/

Windows:

$path = "$env:APPDATA\Clearminutes\models"
if (!(Test-Path $path)) { mkdir $path }
Get-ChildItem $path

#Retry Download

1. Go to Settings → Transcription → Model
2. Find the model that failed
3. Click Download again
4. Watch for "Downloaded" status or green checkmark

If it fails repeatedly:

1. Check all conditions above
2. Restart Clearminutes
3. Try downloading a smaller model first (e.g. base before large-v3)

#Download Model Manually

If in-app download keeps failing, you can download a whisper.cpp compatible model file manually:

1. Download the appropriate .bin file for your model (e.g. ggml-tiny.bin, ggml-base.bin) from a whisper.cpp model source
2. Place the file in the model directory:
   • macOS: ~/Library/Application Support/Clearminutes/models/
   • Windows: %APPDATA%\Clearminutes\models\
3. Restart Clearminutes
4. The model should auto-detect if the filename matches what Clearminutes expects

#Model Corrupt After Download

Downloaded model file is corrupted or won't load.

#Delete and Re-Download

1. Go to Settings → Transcription → Model
2. Click the gear icon next to the model
3. Click Delete
4. Click Download again
5. Wait for completion

#Manual Deletion

If in-app deletion doesn't work:

macOS:

# List models
ls ~/Library/Application\ Support/Clearminutes/models/

# Delete corrupted model (example: tiny.bin)
rm ~/Library/Application\ Support/Clearminutes/models/tiny.bin

Windows:

# Go to models folder
cd "$env:APPDATA\Clearminutes\models"

# List files
dir

# Delete (example: tiny.bin)
del tiny.bin

Then download again in Clearminutes.

#Check File Integrity

If you downloaded manually, verify the file:

macOS:

# Check file size (should match expected)
ls -lh ~/Library/Application\ Support/Clearminutes/models/large-v3.bin
# Should be around 2.9 GB

# Check if readable (no permission errors)
file ~/Library/Application\ Support/Clearminutes/models/large-v3.bin

#Model Not Loading

Model is downloaded, but app can't load it.

#Restart Clearminutes

Sometimes a stale process prevents loading:

1. Quit Clearminutes (Cmd+Q / Ctrl+Q)
2. Wait 5 seconds
3. Reopen Clearminutes
4. Try transcription again

#Check System Logs

1. Go to Settings → Advanced → View Logs
2. Look for errors mentioning "Whisper" or "model load"
3. Common errors:
   • "CUDA out of memory" → see Out of Memory Error below
   • "Model file not found" → verify file exists in models folder
   • "Unsupported GPU" → switch to CPU mode or update GPU drivers

#Verify Model File Exists

macOS:

ls -la ~/Library/Application\ Support/Clearminutes/models/

Windows:

dir "$env:APPDATA\Clearminutes\models\"

Model file should be listed (e.g. base.bin, large-v3.bin). If missing, download again →.

#Try Smaller Model

Large models sometimes fail to load. Test with a smaller model:

1. Go to Settings → Transcription → Model
2. Choose tiny or base
3. Download if needed
4. Try transcription

If smaller model works, larger model may require more GPU memory or system resources.

#Disable GPU and Use CPU

Model may not work with your GPU. Force CPU:

1. Go to Settings → Advanced → GPU
2. Choose CPU Only
3. Restart Clearminutes
4. Try loading model again

If it works on CPU, you have a GPU compatibility issue (see GPU Acceleration →).

#Out of Memory Error

"CUDA out of memory" or "malloc failed" error during transcription.

#Close Other Applications

Free up system RAM:

1. Quit browser (Chrome, Firefox use 2-4 GB per tab)
2. Close video editors, photo apps, IDE
3. Restart computer
4. Try transcription again

Available memory:

• Windows: Open Task Manager (Ctrl+Shift+Esc) → Performance → Memory
• macOS: Open Activity Monitor → Memory tab

#Use Smaller Model

Large models require more memory:

1. Go to Settings → Transcription → Model
2. Choose smaller model:
   • large-v3 (2.9GB needed) → medium (1.4GB)
   • medium → small (460MB)
   • small → base (140MB)
3. Download and try again

#Reduce Recording Length

Very long recordings require more temporary memory:

1. Break into 30-minute segments
2. Transcribe each separately
3. Combine transcripts manually after

#Increase System Swap (Advanced)

If you can't close apps or upgrade RAM, increase virtual memory (slower):

Windows:

1. Right-click This PC → Properties
2. Click Advanced system settings
3. Click Performance Settings
4. Go to Advanced tab
5. Click Change under Virtual Memory
6. Increase to 2x your RAM (e.g. if 8GB RAM, set to 16GB virtual)

macOS:

• Virtual memory is managed automatically by the OS

#GPU Memory Error

"CUDA out of GPU memory" or similar GPU-specific error.

#Close GPU-Heavy Apps

Some apps monopolize GPU memory:

1. Close video games
2. Close photo/video editing software
3. Close streaming software (OBS, etc.)
4. Restart Clearminutes

#Reduce Whisper Model Size

Large models need more GPU memory:

Switch to a smaller model (large-v3 → medium).

#Force CPU Mode

Use CPU transcription instead of GPU:

1. Go to Settings → Advanced → GPU
2. Choose CPU Only
3. Restart Clearminutes

Transcription will be slower but use system RAM instead of GPU VRAM.

#Reduce Model Size (Long-Term)

If you consistently run out of VRAM, use a smaller Whisper model:

• large-v3 → medium
• medium → small
• small → base

#Model File Location and Manual Management

If you need to manually move, backup, or delete models:

#Model Locations

macOS:

~/Library/Application Support/Clearminutes/models/

Windows:

%APPDATA%\Clearminutes\models\

(Typically: C:\Users\[YourUsername]\AppData\Roaming\Clearminutes\models\)

#Backup Models

To avoid re-downloading:

# macOS
cp -r ~/Library/Application\ Support/Clearminutes/models ~/backup_models

# Windows
xcopy "%APPDATA%\Clearminutes\models" C:\backup_models /E /I

#Change Model Location

Model files are stored in the default application data directory. To move them to an external drive, manually copy the model files and update any relevant settings in the app.

#Manually Delete Models

To free up space:

macOS:

# Delete specific model
rm ~/Library/Application\ Support/Clearminutes/models/large-v3.bin

# Delete all models
rm -rf ~/Library/Application\ Support/Clearminutes/models/*

Windows:

# Delete specific model
del "$env:APPDATA\Clearminutes\models\large-v3.bin"

# Delete all models
Remove-Item "$env:APPDATA\Clearminutes\models\*"

Then download fresh in Clearminutes.

#Still Having Issues?

If model errors persist:

1. Collect information:

   • Go to Settings → Advanced → View Logs
   • Screenshot any error messages
   • Note model size and system specs (RAM, GPU)

2. Contact support:

   • https://support.clearminutes.app
   • Include logs, error message, and system info
   • Describe what you've already tried

3. Nuke and rebuild (last resort):

   • Uninstall Clearminutes
   • Delete: ~/Library/Application Support/Clearminutes/ (macOS) or %APPDATA%\Clearminutes\ (Windows)
   • Reinstall Clearminutes
   • Download models fresh

#Related

• Model Selection — Choosing the right Whisper model
• GPU Acceleration — Setting up GPU support
• Privacy Settings — Model storage locations