Speakr
Docker-Anwendung from learnedmachine's Repository
Übersicht
Speakr is a self-hosted AI transcription and intelligent note-taking platform. Transform your audio recordings into organized, searchable, and intelligent notes with speaker identification, AI chat, semantic search, and collaboration features.
Key Features:
- AI-powered transcription with speaker identification
- Voice profiles for automatic speaker recognition
- Interactive chat with your recordings
- Semantic search across all recordings (Inquire Mode)
- Internal sharing and group collaboration
- Smart tagging with custom AI prompts
- Auto-deletion and retention policies
- Automated export to Obsidian/Logseq
- Full internationalization (EN, ES, FR, DE, ZH)
- Light/dark themes with customizable colors
IMPORTANT: Requires API keys for OpenAI/OpenRouter or local AI services for transcription and text generation.
Anforderungen
Before starting, you MUST configure API keys:
1. Create a .env file or use environment variables
2. Set TEXT_MODEL_API_KEY (OpenRouter or OpenAI)
3. Set TRANSCRIPTION_API_KEY (OpenAI Whisper)
4. Set ADMIN_USERNAME and ADMIN_PASSWORD
See the GitHub documentation for complete setup instructions.
Laufzeit-Argumente
- Web-UI
http://[IP]:[PORT:8899]- Netzwerk
bridge- Shell
sh- Privilegiert
- false
Konfiguration der Vorlage
Web interface port
- Ziel
- 8899
- Standard
- 8899
- Wert
- 8899
Storage for uploaded audio files
- Ziel
- /data/uploads
- Standard
- /mnt/user/appdata/speakr/uploads
- Wert
- /mnt/user/appdata/speakr/uploads
Database and application state
- Ziel
- /data/instance
- Standard
- /mnt/user/appdata/speakr/instance
- Wert
- /mnt/user/appdata/speakr/instance
Exported transcriptions (for Obsidian, etc.)
- Ziel
- /data/exports
- Standard
- /mnt/user/appdata/speakr/exports
- Wert
- /mnt/user/appdata/speakr/exports
Watch directory for automatic file processing
- Ziel
- /data/auto-process
- Standard
- /mnt/user/appdata/speakr/auto-process
- Wert
- /mnt/user/appdata/speakr/auto-process
API key for OpenRouter or OpenAI (required for summaries)
- Ziel
- TEXT_MODEL_API_KEY
API endpoint for text generation
- Ziel
- TEXT_MODEL_BASE_URL
- Standard
- https://openrouter.ai/api/v1
- Wert
- https://openrouter.ai/api/v1
Model name for text generation
- Ziel
- TEXT_MODEL_NAME
- Standard
- openai/gpt-4o-mini
- Wert
- openai/gpt-4o-mini
OpenAI API key for Whisper transcription (required)
- Ziel
- TRANSCRIPTION_API_KEY
Whisper API endpoint
- Ziel
- TRANSCRIPTION_BASE_URL
- Standard
- https://api.openai.com/v1
- Wert
- https://api.openai.com/v1
Whisper model to use
- Ziel
- WHISPER_MODEL
- Standard
- whisper-1
- Wert
- whisper-1
Initial admin username
- Ziel
- ADMIN_USERNAME
- Standard
- admin
- Wert
- admin
Initial admin email
- Ziel
- ADMIN_EMAIL
- Standard
- admin@example.com
- Wert
- admin@example.com
Initial admin password (CHANGE THIS!)
- Ziel
- ADMIN_PASSWORD
Allow new user registration (true/false)
- Ziel
- ALLOW_REGISTRATION
- Standard
- false
- Wert
- false
Timezone for date/time display (e.g., America/New_York)
- Ziel
- TIMEZONE
- Standard
- UTC
- Wert
- UTC
Logging level: DEBUG, INFO, WARNING, ERROR
- Ziel
- LOG_LEVEL
- Standard
- INFO
- Wert
- INFO
Maximum tokens for AI summaries
- Ziel
- SUMMARY_MAX_TOKENS
- Standard
- 8000
- Wert
- 8000
Maximum tokens for chat responses
- Ziel
- CHAT_MAX_TOKENS
- Standard
- 5000
- Wert
- 5000
Split large files for API limits (true/false)
- Ziel
- ENABLE_CHUNKING
- Standard
- true
- Wert
- true
Chunk size limit (e.g., 20MB or 1200s)
- Ziel
- CHUNK_LIMIT
- Standard
- 20MB
- Wert
- 20MB
Overlap between chunks in seconds
- Ziel
- CHUNK_OVERLAP_SECONDS
- Standard
- 3
- Wert
- 3
Enable AI-powered semantic search across recordings (true/false)
- Ziel
- ENABLE_INQUIRE_MODE
- Standard
- false
- Wert
- false
Watch directory for automatic file processing (true/false)
- Ziel
- ENABLE_AUTO_PROCESSING
- Standard
- false
- Wert
- false
Processing mode: admin_only, user_directories, or single_user
- Ziel
- AUTO_PROCESS_MODE
- Standard
- admin_only
- Wert
- admin_only
Check interval for new files (seconds)
- Ziel
- AUTO_PROCESS_CHECK_INTERVAL
- Standard
- 30
- Wert
- 30
Automatically export transcriptions to markdown (true/false)
- Ziel
- ENABLE_AUTO_EXPORT
- Standard
- false
- Wert
- false
Include transcription in exports (true/false)
- Ziel
- AUTO_EXPORT_TRANSCRIPTION
- Standard
- true
- Wert
- true
Include summary in exports (true/false)
- Ziel
- AUTO_EXPORT_SUMMARY
- Standard
- true
- Wert
- true
Enable automatic deletion of old recordings (true/false)
- Ziel
- ENABLE_AUTO_DELETION
- Standard
- false
- Wert
- false
Days to retain recordings (0 = disabled)
- Ziel
- GLOBAL_RETENTION_DAYS
- Standard
- 90
- Wert
- 90
Deletion mode: audio_only or full_recording
- Ziel
- DELETION_MODE
- Standard
- audio_only
- Wert
- audio_only
Allow users to delete their own recordings (true/false)
- Ziel
- USERS_CAN_DELETE
- Standard
- true
- Wert
- true
Enable user-to-user sharing (true/false)
- Ziel
- ENABLE_INTERNAL_SHARING
- Standard
- false
- Wert
- false
Display usernames in interface (true/false)
- Ziel
- SHOW_USERNAMES_IN_UI
- Standard
- false
- Wert
- false
Allow public share links (true/false)
- Ziel
- ENABLE_PUBLIC_SHARING
- Standard
- true
- Wert
- true
Database connection string
- Ziel
- SQLALCHEMY_DATABASE_URI
- Standard
- sqlite:////data/instance/transcriptions.db
- Wert
- sqlite:////data/instance/transcriptions.db
Statistik herunterladen
Gesamte Downloads im Laufe der Zeit
Einzelheiten
learnedmachine/speakrFühren Sie Speakr auf Unraid aus.
Speakr ist gelistet in Community Apps für Unraid OS. Erkunden Sie Unraid, um einen flexiblen Heimserver, ein NAS oder ein Heimlabor aufzubauen.