Real-Time Transcription

Watch your words appear as you speak with minimal lag:

• Live Display: Floating overlay shows real-time captions
• Auto-Correction: AI refines transcripts as context becomes clear
• Confidence Scoring: See which words may need manual review
• Instant Search: Find specific topics during live conversations

Speaker Diarization

Automatically identify and label different speakers in conversations:

{
  "transcription": {
    "speakers": [
      {
        "id": "speaker_1",
        "label": "John (You)",
        "voiceprint": "...",
        "color": "#4dabf7"
      },
      {
        "id": "speaker_2",
        "label": "Sarah",
        "voiceprint": "...",
        "color": "#51cf66"
      }
    ],
    "autoDetect": true,
    "maxSpeakers": 10
  }
}

Smart Punctuation

AI adds proper punctuation, capitalization, and formatting:

• Sentence Detection: Automatic periods, commas, question marks
• Capitalization: Proper names, sentence starts, acronyms
• Paragraph Breaks: Detect topic changes and speaker switches
• Numbers & Dates: Convert spoken numbers to digits
• Filler Word Removal: Optionally remove "um", "uh", "like"

Live Captions

Display real-time subtitles on screen:

Transcript History

All transcripts are automatically saved and searchable:

• Unlimited Storage: Keep all transcripts (configurable retention)
• Full-Text Search: Find specific words or phrases instantly
• Date/Time Filtering: Browse by meeting date and time
• Tagging System: Organize transcripts with custom tags
• Favorites: Mark important conversations

Meeting Transcription

Scenario: Capture every detail from Zoom/Teams meetings

Setup:

• Enable transcription for Zoom/Teams in meeting profile
• Turn on speaker detection to identify participants
• Auto-export transcript at meeting end
• Share via email or save to note-taking app

{
  "meetingTranscription": {
    "autoStart": true,
    "speakerDetection": true,
    "removeFillerswords": true,
    "highlightActionItems": true,
    "autoSummarize": true,
    "exportFormat": ["docx", "pdf"]
  }
}

Content Creation

Scenario: Transcribe podcast interviews and YouTube videos

Setup:

• High accuracy mode for clean transcripts
• Speaker labels for host and guests
• Timestamp markers every 30 seconds
• Export as SRT/VTT for video subtitles

Accessibility

Scenario: Live captions for deaf/hard-of-hearing users

Setup:

• Enable floating caption window
• Large font size for readability
• High contrast color scheme
• Fast mode for minimal delay
• Send captions to second monitor

Learning & Education

Scenario: Transcribe lectures and study sessions

Setup:

• Auto-save all transcripts with tags
• Enable keyword highlighting (key terms)
• Create summary notes automatically
• Export to Notion, OneNote, or Evernote

Export Formats

Export transcripts in multiple formats:

• Plain Text (.txt): Simple, compatible everywhere
• Microsoft Word (.docx): Formatted with speakers and timestamps
• PDF: Professional, shareable format
• SRT/VTT: Video subtitle formats for editing software
• JSON: Raw data with metadata for developers
• Markdown: For documentation and note-taking apps

Integration with Note Apps

Automatically send transcripts to your favorite apps:

{
  "integrations": {
    "notion": {
      "enabled": true,
      "autoSend": true,
      "database": "Meeting Notes"
    },
    "onenote": {
      "enabled": true,
      "notebook": "Work Notes",
      "section": "Meetings"
    },
    "obsidian": {
      "enabled": true,
      "vault": "Personal",
      "folder": "Transcripts"
    }
  }
}

API Access

Developers can access transcripts via REST API:

// Fetch recent transcripts
const response = await fetch('http://localhost:8080/api/transcripts', {
  headers: { 'Authorization': 'Bearer YOUR_API_KEY' }
});

const transcripts = await response.json();

// Get specific transcript
const transcript = await fetch(
  'http://localhost:8080/api/transcripts/abc123'
);

const data = await transcript.json();
console.log(data.text);       // Full transcript text
console.log(data.speakers);   // Speaker information
console.log(data.timestamps); // Word-level timestamps

Data Storage

• Encrypted Database: Transcripts stored in encrypted SQLite database
• Local Only: No cloud storage or external servers
• Configurable Retention: Auto-delete transcripts after X days
• Manual Deletion: Delete individual transcripts anytime

Security Features

{
  "privacy": {
    "encryptTranscripts": true,
    "encryptionKey": "user-password-derived",
    "retentionDays": 90,           // Auto-delete after 90 days
    "excludeApps": ["signal.exe"], // Never transcribe these apps
    "pauseDuringPrivacy": true,    // Stop when screen locked
    "redactSensitive": true        // Hide credit cards, SSNs, etc.
  }
}

Transcription Not Starting

Solution:

1. Check Feature Status

   Verify transcription is enabled in Settings → Features → Transcription

2. Verify AI Model Download

   Ensure AI model is fully downloaded (Settings shows "Ready" status)

3. Check Microphone Input

   Confirm audio is being received in Jyv Desktop's audio meter

4. Restart Transcription Engine

   Settings → Transcription → Restart Engine

Poor Transcription Accuracy

Cause: Background noise, poor microphone, or wrong language

Solution:

• Enable Accurate Mode for better quality
• Ensure Noise Suppression is active
• Verify correct language is selected
• Use better microphone or improve room acoustics
• Speak clearly and at consistent volume

Speaker Detection Not Working

Solution:

• Train your voiceprint in transcription settings
• Increase minimum speaker detection confidence
• Manually label speakers after transcription
• Ensure speakers have distinct voices (not too similar)
• Reduce background noise for better voice separation

High CPU Usage

Solution:

• Switch from Accurate to Fast or Balanced mode
• Enable GPU Acceleration if available
• Disable speaker detection if not needed
• Close other resource-intensive applications
• Reduce transcription sample rate (24kHz instead of 48kHz)