akouo-router

akouo-router

Purpose

akouo-router is the meta-ear of akoúō. It does not analyze the sound directly. It analyzes the listening situation and chooses which listening modes should be used.

The router keeps the system from treating every sonic object as the same kind of problem. It decides whether the next response should remain descriptive, technical, forensic, ecological, critical, speculative, or comparative.

The router also checks conceptual confusion: signal versus meaning, music versus genre label, voice versus transcript, speech versus identity, caption versus sound, audiovisual synchronization versus causality, access versus normal-hearing assumption, material vibration versus measured evidence, soundscape versus acoustemology versus aurality, machine listening versus voice agent, archive versus evidence, affect versus emotion, and speculation versus proof.

When To Use

Use this skill whenever an agent is asked to:

• listen to a sound, audio file, recording, prompt, transcript, field note, spectrogram, archive fragment, dataset, or sonic concept
• analyze or interpret audio without a clearly specified mode
• choose between technical, perceptual, musical/aesthetic, affective, voice/speech, audiovisual/scenic, accessibility, material/event, archival, ecological, political, or fictional listening
• prevent overreach in a sensitive or uncertain listening situation
• recommend a listening chain for another command or agentic workflow

Core Question

What kind of listening does this situation require?

Routing Refinements

• Route by evidence first: audio file, metadata, waveform, spectrogram, transcript, prompt, field note, archive note, model output, or mixed input do not support the same claims.
• Route by risk second: forensic, legal, political, colonial, identity, accessibility, surveillance, and platform contexts need corrective caution.
• Route by conceptual frame third: technical measurement, perceptual morphology, musical/aesthetic organization, bodily affect, voice/speech, audiovisual scene, accessibility and hearing norm, material event, transduction, testimony, ecological relation, political mediation, memory and lineage, or declared fiction.
• Add a corrective mode whenever the likely primary mode could overreach, aestheticize harm, erase mediation, or convert speculation into evidence.
• Prefer multi-mode commands when the user asks for a complete reading or when the object crosses evidence, mediation, politics, and fiction.
• For autonomous workflows, return enough evidence limits, forbidden assumptions, and mode-chain logic that another app or agent can execute the route without inventing missing evidence.
• Correct for vococentrism: the presence of speech pulls routing toward voice modes by default, but a task about ambience, music, habitat, or material behavior keeps its own primary even when someone is speaking in the recording.
• Never let one frame capture the whole route: keep at least one alternative reading alive (route_reasoning for router output, alternative_reading in mode outputs), especially when confidence is high.