sonarwise

Pluggable audio perception engine. Hear. Search. Retrieve.

sonarwise indexes any audio like meetings, calls, podcasts, factory floors, lectures and makes it searchable by text, speaker, sound events, or audio similarity. Every component is pluggable: swap transcription, embedding, diarization, or storage without changing your code.

Features

• Transcription : Whisper, Faster Whisper, or bring your own ASR
• Audio Embeddings : CLAP joint text-audio space for semantic search
• Speaker Diarization : know who said what (pyannote)
• Speaker Registry : track speakers across files by voiceprint
• Audio Event Detection : detect alarms, machinery, glass breaks, and 50+ sound types
• Live Streaming : real-time transcription with keyword and event callbacks
• Export : SRT, VTT, JSON, CSV, TXT, meeting notes
• Pluggable Architecture : every component swappable via base classes
• CLI : full command-line interface

Install

# Core (no ML dependencies)
pip install sonarwise

# With all features
pip install sonarwise[all]

# Pick what you need
pip install sonarwise[whisper]          # Whisper transcription
pip install sonarwise[faster-whisper]   # Faster Whisper (CTranslate2)
pip install sonarwise[clap]            # CLAP audio embeddings
pip install sonarwise[diarization]     # Speaker diarization
pip install sonarwise[speaker]         # Speaker identification
pip install sonarwise[events]          # Audio event detection
pip install sonarwise[live]            # Live microphone capture

Requires: ffmpeg (sudo apt install ffmpeg or brew install ffmpeg)

Quick Start

from sonarwise import SonarWise

sw = SonarWise(diarization=True, events=True)

# Index audio
sw.index("meeting.wav")
sw.index_folder("./recordings/")

# Search by text
results = sw.query("budget discussion", top_k=5)
for r in results:
    print(f"[{r.speaker_name}] {r.transcript} (score: {r.score})")

# Search by audio similarity
results = sw.query_audio("alarm_clip.wav", top_k=5)

# Search by speaker
results = sw.query("budget", speaker="Ant", top_k=5)

# Search by event
results = sw.query_events(event="machine_fault", top_k=5)

Speaker Intelligence

# Register a speaker
sw.register_speaker("Ant", reference_audio="ant_voice.wav")

# Query by speaker
results = sw.query("budget", speaker="Ant")

# Speaker timeline
timeline = sw.speaker_timeline("meeting.wav")

# Speaker stats
stats = sw.speaker_stats("meeting.wav")

# Find speaker across files
presence = sw.find_speaker_across(speaker="Ant", folders=["./meetings/"])

Live Mode

sw = SonarWise(mode="live", diarization=True, events=True)

@sw.on("transcript")
def on_speech(segment):
    print(f"[{segment.speaker_name}] {segment.transcript}")

@sw.on("keyword", words=["budget", "deadline", "risk"])
def on_keyword(segment):
    send_alert(segment.transcript)

@sw.on("sound_event", events=["alarm", "glass_break"])
def on_danger(event):
    trigger_alert(event)

sw.listen(source="microphone")

Plug Any Model

Every component is swappable:

from sonarwise import SonarWise
from sonarwise.core.transcriber import BaseTranscriber

class MyTranscriber(BaseTranscriber):
    def transcribe(self, audio):
        return my_model.process(audio)

sw = SonarWise(transcriber=MyTranscriber())

Pluggable slots:

CLI

sonarwise index meeting.wav
sonarwise query "budget discussion"
sonarwise speakers meeting.wav
sonarwise timeline meeting.wav
sonarwise listen --source microphone --on-keyword "budget,risk"
sonarwise export meeting.wav --format srt
sonarwise stats

Export

sw.export("meeting.wav", format="srt", output="subtitles.srt")
sw.export("meeting.wav", format="json", output="segments.json")
sw.export("meeting.wav", format="txt", output="transcript.txt")
sw.export("meeting.wav", format="notes", output="meeting_notes.md")

Ecosystem

sonarwise is part of the VK-Ant AI perception ecosystem:

from sightrag import SightRAG
from sonarwise import SonarWise
from adaptive_intelligence import AdaptiveRAG

brain = AdaptiveRAG()
brain.register_source("visual", SightRAG())
brain.register_source("audio", SonarWise())

# One query hits both eyes and ears
results = brain.query("when did machine 3 start failing?")

License

Apache 2.0 License. Built by Venkatkumar Rajan.