How vk-ant/sonarwise works
Sonarwise is a pluggable audio indexing and retrieval engine. Users submit recordings through the Python API or CLI; the central pipeline prepares and segments audio, derives transcripts and optional speaker/event/language metadata, stores searchable segments, and serves text, audio, speaker, and event queries. It also documents live listening, conversation analytics, clip extraction, and export. The pipeline excerpt is partial, so the graph reflects documented capabilities and indexed module references without asserting unshown implementation details.
Main entry points: pipeline.py SonarWise pipeline · store.py Segment and speaker store · __init__.py Python API · cli.py Command line
Components 20
User entry points
- Python API
sonarwise/__init__.py - Command line
sonarwise/cli.py
Audio indexing
- SonarWise pipeline
sonarwise/core/pipeline.py - Audio I/O
sonarwise/utils/audio_io.py - Audio preprocessing
sonarwise/core/preprocessor.py - Audio chunking
sonarwise/core/chunker.py - Transcription
sonarwise/core/transcriber.py - Speaker diarization
sonarwise/core/diarizer.py - Event classification
sonarwise/core/event_classifier.py - Language detection
sonarwise/core/language_detector.py - Live listening
sonarwise/core/stream_listener.py
Search and intelligence
- Audio embeddings
sonarwise/core/embedder.py - Speaker voiceprints
sonarwise/core/speaker_embedder.py - Segment and speaker store
sonarwise/core/store.py - Semantic and event search
sonarwise/core/pipeline.py - Conversation analytics
sonarwise/core/pipeline.py - Clip extraction and export
sonarwise/core/pipeline.py
Runtime services
- Audio and result models
sonarwise/core/models.py - Configuration
sonarwise/config.py
Other components
- Audio user
Connections 24
- Audio user → Python API: uses
README.md - Audio user → Command line: runs commands
README.md - Python API → SonarWise pipeline: creates
sonarwise/__init__.py - Command line → SonarWise pipeline: invokes
sonarwise/cli.py - SonarWise pipeline → Configuration: configures
sonarwise/core/pipeline.py - SonarWise pipeline → Audio I/O: loads audio
sonarwise/core/pipeline.py - SonarWise pipeline → Audio preprocessing: preprocesses
sonarwise/core/pipeline.py - SonarWise pipeline → Audio chunking: segments audio
sonarwise/core/pipeline.py - SonarWise pipeline → Transcription: transcribes segments
sonarwise/core/pipeline.py - SonarWise pipeline → Speaker diarization: assigns speakers
sonarwise/core/pipeline.py - SonarWise pipeline → Event classification: classifies events
sonarwise/core/pipeline.py - SonarWise pipeline → Language detection: detects language
sonarwise/core/pipeline.py - SonarWise pipeline → Audio embeddings: embeds audio
sonarwise/core/pipeline.py - SonarWise pipeline → Speaker voiceprints: embeds voices
sonarwise/core/pipeline.py - SonarWise pipeline → Segment and speaker store: stores segments
sonarwise/core/pipeline.py - SonarWise pipeline → Audio and result models: uses data models
sonarwise/core/pipeline.py - SonarWise pipeline → Semantic and event search: serves queries
README.md - Semantic and event search → Segment and speaker store: searches segments
sonarwise/core/pipeline.py - Semantic and event search → Audio embeddings: embeds text queries
sonarwise/core/pipeline.py - SonarWise pipeline → Conversation analytics: analyzes conversations
sonarwise/core/pipeline.py - Conversation analytics → Segment and speaker store: reads segments
sonarwise/core/pipeline.py - SonarWise pipeline → Clip extraction and export: extracts and exports
sonarwise/core/pipeline.py - SonarWise pipeline → Live listening: starts listening
sonarwise/core/pipeline.py - Live listening → Audio I/O: captures audio
sonarwise/core/stream_listener.py