chore(chat): remove deployed-chat voice mode

[waleedlatif1]

Summary

• Removes the voice-first interface and TTS playback from the deployed chat. Workspace dictation is a separate feature and is kept.
• Deletes the VoiceInterface UI + particles canvas, the chat mic input, the TTS audio-streaming hook, the /api/proxy/tts/stream relay and its contract, and the voice-settings query hook.
• Unpicks the voice wiring in chat.tsx and use-chat-streaming: audio stream handler, sentence-splitting for speech, voice-first mode state, and the isVoiceInput plumbing through ChatInput.
• Drops the now-dead chatId branch from /api/speech/token, which leaves resolveDeployedChatCaller unused, so that goes too.
• Removes what the above orphaned: MAX_CHAT_SESSION_MS, the noop util, write-only refs in use-chat-streaming, and the chatId field on the speech contract.

Notes for review

• Security-relevant: the chatId branch in /api/speech/token was the anonymous public-chat path. With chat voice gone it had no legitimate caller, but leaving it would have kept an unauthenticated STT relay spending the platform ElevenLabs key. Removed rather than left dormant.
• Kept deliberately: /api/settings/voice, lib/speech/* and use-speech-to-text — the workspace home input still uses them for dictation.
• Kept deliberately: the voice-output usage source, its pg enum value and its label. Postgres cannot drop an enum value, and historical usage_log rows written while the TTS relay was live still need a label to render on the credit-usage page.
• This removes the route secured in fix(security): meter and throttle the deployed-chat TTS relay #6212. That fix shipped first, so staging was never left exposed.

Type of Change

• Chore / removal

Testing

555 tests pass across chat, speech, settings, usage-logs and billing. Typecheck clean on apps/sim and packages/db; all 23 lint tasks pass. Removed two /api/speech/token tests and one use-chat-streaming test that covered the deleted paths. Not manually exercised in a browser.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		Aug 3, 2026 7:00pm

[cursor]

PR Summary

Medium Risk
Large deletion touching chat UX and removes an anonymous STT/TTS surface, but core text chat and workspace dictation remain; regression risk is mainly deployed-chat messaging/streaming, not auth bypass.

Overview
Removes voice-first mode and TTS playback from the deployed chat UI. Visitors now get text-only chat: no full-screen VoiceInterface, mic controls on ChatInput, or auto-played assistant audio while streaming.

Client: Deletes voice UI (VoiceInterface, particles, VoiceInput), useAudioStreaming and its tests, and the useVoiceSettings query. chat.tsx drops voice-first state, TTS wiring, and the isVoiceInput path through send/stream handling. useChatStreaming no longer accepts voice settings or sentence-splitting TTS callbacks.

API / contracts: Removes /api/proxy/tts/stream (route, tests, tts-stream contract) and the anonymous chatId STT path on /api/speech/token, plus unused resolveDeployedChatCaller. Speech tokens are session + verified workspaceId only (editor dictation path).

Intentionally kept: Workspace home dictation (useSpeechToText, /api/speech/token, lib/speech/*, voice settings API) and historical voice-output usage labels for past billing rows.

^{Reviewed by Cursor Bugbot for commit b5207dd. Configure here.}

[greptile-apps]

Greptile Summary

Removes deployed-chat voice input and text-to-speech while preserving workspace dictation.

• Deletes the deployed voice interface, audio streaming hook, TTS proxy route, and associated contracts and tests.
• Simplifies deployed-chat input and response streaming to text and attachments.
• Removes the anonymous deployed-chat branch from speech-token issuance while retaining session-authenticated, workspace-scoped token issuance.

Confidence Score: 5/5

The PR appears safe to merge with no actionable changed-code defects identified.

The removed voice-only UI, streaming, API, and contract paths have no remaining callers, while text chat, attachment submission, and authenticated workspace dictation retain consistent call and authorization contracts.

Important Files Changed

Filename	Overview
apps/sim/app/(interfaces)/chat/[identifier]/chat.tsx	Removes deployed-chat voice state, controls, audio playback, and voice-specific submission wiring while preserving text and attachment submission.
apps/sim/app/(interfaces)/chat/components/input/input.tsx	Removes voice-only properties and microphone controls and consistently updates the submission callback to pass attachments as its second argument.
apps/sim/app/(interfaces)/chat/hooks/use-chat-streaming.ts	Removes voice-specific sentence segmentation and audio callbacks without altering the established text-stream parsing and terminal-state handling.
apps/sim/app/api/speech/token/route.ts	Removes the intentionally deprecated anonymous deployed-chat token branch and retains authenticated workspace membership, billing, rate-limit, and usage checks.
apps/sim/lib/api/contracts/media/speech.ts	Removes the obsolete chatId request field while preserving the workspaceId contract used by workspace dictation.
apps/sim/app/api/proxy/tts/stream/route.ts	Deletes the deployed-chat TTS relay as part of the voice-output feature removal.

_{Reviews (1): Last reviewed commit: "chore(chat): remove deployed-chat voice ..." | Re-trigger Greptile}