docs: add voice-to-text chat input design

Design document for adding microphone button with Whisper API
transcription to the ChatInput component.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

docs/plans/2026-01-28-voice-to-text-chat-input-design.md

@@ -0,0 +1,105 @@
+# Voice-to-Text Chat Input Feature
+
+## Overview
+
+Add a microphone button to the ChatInput component that allows users to record voice and transcribe it to text using OpenAI's Whisper API, similar to ChatGPT.
+
+## Requirements
+
+- Toggle recording with click (click to start, click to stop)
+- Space key triggers recording toggle when input is focused
+- Maximum recording duration: 2 minutes
+- Use OpenAI Whisper API for transcription
+- Frontend API route handles the Whisper call
+
+## Architecture
+
+### Components to Create/Modify
+
+1. **New API route** - `/api/transcribe/route.ts`
+   - Accepts audio blob (webm) via POST
+   - Calls OpenAI Whisper API (`whisper-1` model)
+   - Returns transcribed text
+
+2. **New hook** - `useVoiceRecording.ts`
+   - Manages MediaRecorder state
+   - Handles start/stop recording
+   - Enforces 2-minute max duration
+   - Sends audio to transcribe API
+
+3. **Modified component** - `ChatInput.tsx`
+   - Add mic button (left of send button)
+   - Visual feedback during recording
+   - Space key handler for toggle
+
+### Environment Variable
+
+- `OPENAI_API_KEY` (server-side only)
+
+## Hook Design: `useVoiceRecording.ts`
+
+### State
+- `isRecording` - boolean for recording state
+- `isTranscribing` - boolean for API call in progress
+- `error` - string for error messages
+
+### Functions
+- `startRecording()` - Request mic permission, start MediaRecorder
+- `stopRecording()` - Stop recording, send to API, return text
+- `toggleRecording()` - Convenience function for button/space key
+
+### Constraints
+- Auto-stop at 2 minutes with timer
+- Show elapsed time during recording
+- Cleanup on unmount
+
+## UI/UX Design
+
+### Mic Button Placement
+- Position: Left of the send button, inside the input wrapper
+- Icon: Microphone (Phosphor icons)
+
+### Visual States
+
+| State | Mic Button | Input Area |
+|-------|-----------|------------|
+| Idle | Gray mic icon | Normal |
+| Recording | Red pulsing mic + elapsed time | Red border glow |
+| Transcribing | Spinner | Disabled, "Transcribing..." placeholder |
+| Error | Mic with warning | Inline error |
+
+### Button Behavior
+- Visible when input is empty or has text
+- Hidden during streaming (when stop button shows)
+
+### Accessibility
+- `aria-label` updates based on state
+- Screen reader announcements
+- Space key to toggle when focused
+
+## API Route: `POST /api/transcribe`
+
+### Request
+- Content-Type: `multipart/form-data`
+- Body: `audio` file (webm blob)
+
+### Response
+```json
+{ "text": "transcribed text here" }
+```
+
+### Error Responses
+- `400` - No audio file provided
+- `401` - Missing API key configuration
+- `413` - File too large (> 25MB)
+- `500` - Whisper API error
+
+## File Changes Summary
+
+| File | Action |
+|------|--------|
+| `src/app/api/transcribe/route.ts` | Create |
+| `src/components/contextual/Chat/components/ChatInput/useVoiceRecording.ts` | Create |
+| `src/components/contextual/Chat/components/ChatInput/ChatInput.tsx` | Modify |
+| `src/components/contextual/Chat/components/ChatInput/useChatInput.ts` | Minor modify (space key coordination) |
+| `.env.default` | Add OPENAI_API_KEY placeholder |