Skip to content

feat: Broadcast Voice (Clarity) #1

Description

@ivalsaraj

Goal

Add a Broadcast Voice control (Off / Low / Medium / High) that makes the voice sound clearer and more "present/broadcast" — a gentle presence lift coupled with an automatic de-esser — without changing the identity of the original voice.

Architecture

Two new allocation-free DSP stages added to the existing VoiceChain: a wide-Q presence peaking bell (Biquad.setPeaking) and a subtractive split-band DeEsser (Dynamics.swift). They are driven by a single ClarityLevel enum carried on VoiceChainSettings and injected by AudioModel on top of the active preset (independent of the noise preset and of Voice Polish). The presence boost is followed by the de-esser so added "air" can never become harsh sibilance; the existing limiter (already last in the chain) catches any peaks. When clarity == .off, presence is bypass and the de-esser is a perfect identity, so every existing preset behaves byte-for-byte as before.

Plan

docs/plans/2026-06-15-broadcast-voice-clarity.md

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

No labels
No labels

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions