Future Feature Requests

Status as of 2026-09-27. Done names the pull request. Partly says which part exists. Deferred gives the reason; an entry without a status line is open.

Transcription and Language

• Add multilingual live translation pipeline for transcript lines (German/English first). — Deferred: needs a translation model on the core or a paid online service; neither is chosen.
• Add per-channel custom phrase dictionaries and shortcut expansions. — Deferred: a Vosk grammar narrows recognition to the listed words and silences everything else; needs trials with real show audio first.
• Add confidence scoring and low-confidence highlighting for transcript lines. — Deferred: thresholds can only be set against real recordings; CI has no model.
• Add transcript recording sessions with searchable timeline and export to JSON/SRT/TXT. — Done (#31): transcript log of 5,000 lines, search, export TXT/SRT/JSON, Clear transcript starts a new session.

Automation and Integrations

• Add keyword rule engine with actions (OSC, MIDI, webhook, local command). — Done for OSC and webhook (#33). No MIDI (native module on every host; bridge from OSC instead) and no local command (a transcript line must not choose what runs on the core).
• Add rule presets for show-caller, FOH, monitor engineer, and stage manager workflows. — Deferred: a preset needs the real target's OSC addresses (console, playback); invented ones would fire the wrong cue.
• Add bidirectional OSC status feedback into UI indicators.
• Add optional external STT gateway mode for dedicated transcription hosts.

Operator UX

• Add dedicated transcript view with channel filters, pinned channels, and unread markers. — Done (#31; the view and filters existed, pinning and unread counts added).
• Add type-to-speak with configurable TTS voice and channel routing.
• Add global keyboard macro profiles (per production profile).
• Add transcript bookmarks and incident markers for post-show reports. — Done (#31): bookmarks with a note in the transcript timeline, included in every export.

Reliability and Ops

• Add transcription health panel (model status, CPU load, dropped frames, latency). — Partly: model status and the reason transcription is off are shown in the transcript view; load, dropped frames and latency are open.
• Add offline model manager with one-click language model download/update. — Partly: one-click install of the small English and German Vosk models; updates and other languages are open.
• Add back-pressure and adaptive chunking for low-bandwidth links. — Deferred: needs measurements on a real weak link.
• Add role-based permissions for who may enable transcription per channel. — Done: per-user transcriptionChannelIds in the user permissions.

Hardware and Routing

• Add input routing bridge for external audio interfaces and virtual audio devices. — Needs hardware to build and test against.
• Add optional per-channel AGC/noise suppression stage before STT. — Deferred: whether it helps recognition can only be judged on real audio.
• Add multi-device speaker diarization for shared channels.
• Add configurable retention policy for transcript data and event logs. — Partly: fixed limits (300 events, 5,000 transcript lines since #31) and manual clearing; configurable retention is open.

DECT Client Roadmap

Needs DECT hardware (base stations, beltpacks) for every item below; nothing here can be built or tested against the simulator alone. - Implement real DECT beltpack client stack (registration, heartbeat, channel sync). - Add DECT antenna roaming and handover behavior testing. - Add battery and charging telemetry from physical DECT clients. - Add RF quality visualization and alarm thresholds in dashboard.

Web and Phone Client Backlog

• Add PWA packaging for the Phone Client (home screen install, wake-lock, offline shell). — Done (#32).
• Add push-to-talk lock mode optimized for touchscreen operation. — Done: slide the PTT bar to the end to latch it (PttSlider).
• Add one-tap role presets (FOH, Stage, Camera, Director) for web clients. — Partly: onboarding picks a user, whose profile sets the slots; named role presets are open.
• Add admin option to provision/revoke web clients as managed devices. — Done (#34): invite link and QR code provision; revoke and restore in the Device Manager.

Audio Plugin and Processing Strategy

Deferred as a group: these are alternatives for where plugins run. The Plugin Bridge exists; which of the models below follows is a decision for the product owner, not a task. - Support per-end-device plugin chains (local browser/native host per client). - Support central plugin host with multi-client routing and reusable chain presets. - Add hybrid mode: edge preprocessing (client) plus central mastering chain. - Add plugin capability negotiation (which client can host which format/CPU budget).

Ideas Collected From This Chat

• Add dedicated DECT hardware client implementation after web/phone stabilization. — Needs DECT hardware, see above.
• Keep ProdCom-like features as optional overlays (translation, automation, typed reply). — Held: transcription, bookmarks and keyword rules are all off until configured.
• Add OSC/MIDI keyword triggers with channel-scoped automation rules. — Done for OSC (#33), see keyword rule engine.
• Add transcript-to-action incidents and operator bookmark workflows. — Done (#31 bookmarks, #33 actions).