Settings & Configuration

This comprehensive guide covers all of DeepTalk’s configuration options, helping you customize the application for optimal performance and user experience.

Settings Overview

Accessing Settings

Location: Click the ⚙️ Settings tab in the main navigation

Settings Organization:

Transcription: Speech-to-text service configuration
Processing: AI analysis and enhancement settings
Chat: Conversation and interaction settings
Prompts: AI prompt customization and management
General: Application preferences and data management

Settings Philosophy

DeepTalk is designed to work well with default settings while offering extensive customization for power users. You can:

Start simple: Use defaults and adjust as needed
Customize gradually: Modify settings as you learn the application
Reset anytime: Return to defaults if changes don’t work out
Export/import: Share configurations across installations

Transcription Settings

Service Configuration

Speech-to-Text Service URL:

Default: http://localhost:8000 (for local Speaches installation)
Purpose: Connect to external transcription service for enhanced quality
Options: Local service, cloud service, or disabled (basic transcription only)

Model Selection:

Default model: Systran/faster-distil-whisper-small.en
Performance: Smaller models = faster processing, larger = better accuracy
Language support: Choose models optimized for your primary language
Custom models: Configure custom or specialized transcription models

Service Testing:

Test connection: Verify service availability and response
Model availability: Check which models are available on your service
Performance testing: Measure speed and accuracy with sample content

Audio Processing

Chunk Size Configuration:

Default: 60 seconds (1 minute)
Range: 30 seconds to 5 minutes
Purpose: Split long audio files for better processing
Optimization: Shorter chunks = faster processing, longer = better context

Audio Quality Settings:

Sample rate: Automatic detection and optimization
Bitrate handling: Automatic conversion for optimal processing
Format conversion: Automatic conversion between supported formats
Noise reduction: Basic noise filtering (if available)

Transcription Enhancement

Quality Optimization:

Enable enhanced processing: Use advanced algorithms for better accuracy
Speaker diarization: Attempt to identify different speakers automatically
Punctuation restoration: Add punctuation to improve readability
Capitalization correction: Proper capitalization for names and sentences

Language and Locale:

Primary language: Main language for transcription optimization
Regional settings: Locale-specific formatting and conventions
Multi-language support: Handle multiple languages in single recordings
Custom vocabulary: Add domain-specific terms for better accuracy

Processing Settings

AI Analysis Service

Service Configuration:

AI Analysis URL: Default http://localhost:11434 (for local Ollama)
Model selection: Choose AI model for analysis (llama2, mistral, etc.)
Connection testing: Verify AI service availability and performance
Fallback options: Configure behavior when AI service is unavailable

Analysis Preferences:

Analysis depth: Quick overview vs. comprehensive analysis
Focus areas: Emphasize summaries, action items, sentiment, etc.
Custom prompts: Configure specialized analysis for your use cases
Batch processing: Settings for analyzing multiple transcripts

Transcript Enhancement

Validation and Correction:

Enable transcript validation: Use AI to correct spelling and grammar
Correction types: Spelling, grammar, punctuation, capitalization
Validation strength: Conservative vs. aggressive correction
Manual review: Require human approval for AI corrections

Speaker Enhancement:

Automatic speaker tagging: Add speaker labels automatically
Speaker identification: Attempt to identify speakers by voice
Speaker naming: Convert “Speaker 1” to meaningful names
Consistency enforcement: Maintain speaker names across sessions

Content Processing:

Duplicate removal: Eliminate repeated sentences or phrases
Analyze validated transcript: Use corrected text for AI analysis
One-task-at-a-time: Process analysis tasks separately for better accuracy
Quality thresholds: Minimum confidence levels for processing

Performance and Resource Management

Processing Priority:

Concurrent processing: Number of simultaneous processing tasks
Resource allocation: CPU and memory usage limits
Queue management: Priority handling for different types of content
Background processing: Continue processing when application is minimized

Storage and Caching:

Cache management: Temporary file storage and cleanup
Storage optimization: Compress older transcripts and analysis
Backup frequency: Automatic backup scheduling
Cleanup rules: Automatic removal of temporary and outdated files

Chat Settings

Conversation Modes

Mode Selection:

Vector Search Only: Return relevant excerpts without AI interpretation
RAG Mode: Retrieve relevant content and send to AI for analysis (recommended)
Direct LLM Mode: Send full transcript directly to AI for comprehensive analysis

Mode-Specific Settings:

Vector Search Configuration:

Number of excerpts: How many relevant sections to return (1-10)
Relevance threshold: Minimum similarity score for results
Excerpt length: Size of returned text segments
Context overlap: Overlap between adjacent excerpts

RAG Mode Settings:

Context chunks: Number of relevant sections to include (1-10)
Chunk size: Maximum size of each context chunk
Response optimization: Balance between speed and comprehensiveness
Memory integration: Use conversation history in responses

Direct LLM Settings:

Context limit: Maximum characters to send to AI (2k-16k)
Truncation strategy: How to handle oversized transcripts
Processing timeout: Maximum time to wait for AI response
Fallback behavior: What to do if processing fails

Advanced Chat Configuration

Chunking Strategy:

Method: Speaker-based, time-based, or hybrid chunking
Maximum chunk size: 30 seconds to 3 minutes
Chunk overlap: 0-30 seconds for continuity
Boundary detection: Smart splitting at natural conversation breaks

Memory and Context:

Conversation memory limit: Number of messages to remember (5-50)
Memory compaction: Automatic summarization of old conversations
Context preservation: Maintain important information across sessions
Session management: Handle multiple concurrent conversations

Response Optimization:

Response style: Formal, conversational, or technical tone
Detail level: Brief summaries vs. comprehensive analysis
Citation format: How to reference transcript sources
Accuracy emphasis: Prioritize factual accuracy vs. natural language

General Settings

Application Preferences

Interface and Appearance:

Theme selection: Light, dark, or system theme
Interface density: Comfortable, compact, or custom spacing
Default page: Which page opens when starting DeepTalk
Panel preferences: Default visibility and layout of interface panels

Language and Localization:

Interface language: Application menu and interface language
Date format: Regional date and time formatting preferences
Number format: Decimal and thousands separator preferences
Timezone handling: How to display and interpret timestamps

Data Management

Storage Configuration:

Database location: Where DeepTalk stores your data
Storage limits: Maximum storage usage before cleanup
Backup settings: Automatic backup frequency and location
Import/export: Data portability and migration options

Backup and Recovery:

Automatic backup: Enable/disable scheduled backups
Backup frequency: Daily, weekly, or monthly backups
Backup location: Local, network, or cloud storage
Recovery options: Restore from backup procedures

Privacy and Security:

Data retention: How long to keep different types of data
Local processing: Ensure all processing stays on your machine
External services: Configure which external services to use
Audit logging: Track access and changes to your data

Performance Optimization

System Resource Management:

Memory usage: Limits on RAM usage for processing
CPU allocation: How much processing power DeepTalk can use
Disk space: Monitoring and cleanup of storage usage
Network usage: Bandwidth limits for external service calls

Application Behavior:

Startup options: What to do when launching DeepTalk
Auto-updates: Automatic vs. manual application updates
Error reporting: Anonymous error reporting to help improve DeepTalk
Telemetry: Usage analytics for application improvement

Prompts and AI Customization

Prompt Management

Default Prompts:

System prompts: Core instructions for AI interactions
Analysis prompts: Templates for different types of analysis
Chat prompts: Instructions for conversation interactions
Validation prompts: Guidelines for transcript correction

Custom Prompt Creation:

Prompt templates: Create reusable prompt structures
Variable substitution: Dynamic content insertion in prompts
Prompt testing: Validate prompt effectiveness with sample content
Version control: Track changes and improvements to prompts

Prompt Categories:

Chat prompts: For interactive conversations with transcripts
Analysis prompts: For automated content analysis
Speaker prompts: For speaker identification and tagging
Validation prompts: For transcript correction and enhancement

AI Model Configuration

Model Selection:

Primary model: Main AI model for analysis and chat
Fallback models: Alternative models if primary is unavailable
Specialized models: Different models for different types of content
Model switching: Automatic or manual model selection

Model Parameters:

Temperature: Creativity vs. consistency in AI responses
Response length: Preferred length of AI-generated content
Context window: How much context to provide to the AI
Processing timeout: Maximum time to wait for AI responses

Configuration Management

Settings Backup and Sync

Export Settings:

Full configuration: All settings and customizations
Selective export: Choose specific setting categories
Template creation: Save configurations as templates
Sharing options: Share configurations with team members

Import and Migration:

Configuration import: Apply settings from exported files
Migration tools: Transfer settings between installations
Version compatibility: Handle settings from different DeepTalk versions
Conflict resolution: Handle conflicts between imported and existing settings

Team Configuration

Organizational Settings:

Default configurations: Standard settings for new team members
Mandatory settings: Required configurations for compliance
Recommended settings: Best practice configurations for your organization
Policy enforcement: Ensure consistent configurations across team

Configuration Templates:

Role-based templates: Different configurations for different roles
Project templates: Settings optimized for specific types of projects
Client templates: Configurations for different client requirements
Use case templates: Settings for specific workflows or industries

Troubleshooting Configuration Issues

Common Configuration Problems

Service Connection Issues:

URL verification: Ensure service URLs are correct and accessible
Network connectivity: Check firewall and network settings
Service status: Verify external services are running and responsive
Authentication: Check if services require API keys or credentials

Performance Issues:

Resource allocation: Adjust memory and CPU limits
Processing queues: Optimize concurrent processing settings
Storage management: Clean up temporary files and optimize storage
Network optimization: Adjust settings for your network conditions

Quality Issues:

Service optimization: Fine-tune transcription and AI service settings
Model selection: Choose appropriate models for your content type
Processing parameters: Adjust analysis depth and focus areas
Validation settings: Optimize correction and enhancement settings

Getting Configuration Help

Built-in Help:

Setting descriptions: Detailed explanations for each configuration option
Recommended values: Suggested settings for different use cases
Testing tools: Built-in tools to test and validate configurations
Reset options: Easy ways to return to working configurations

External Resources:

Documentation: Detailed guides for service setup and optimization
Community support: User forums and discussion groups
Technical support: Direct support for configuration assistance
Best practices: Guides for optimal configuration in different scenarios

Next: Explore detailed features → or learn about common workflows →