voice-settingslisted

# VoiceBridge Settings Use `mcp__voicebridge__voice_status`, `mcp__voicebridge__voice_models`, `mcp__voicebridge__voice_configure`, and `mcp__voicebridge__voice_stop` to change the saved speech-model pair. This workflow is configuration only: do not start a voice conversation or use the microphone. 1. Call `mcp__voicebridge__voice_status`. 2. If `ready` is true, explain that a model change ends the current Voice Code session. Ask for confirmation with Codex's structured question UI. If the user confirms, call `mcp__voicebridge__voice_stop`; otherwise end without changing settings. 3. Call `mcp__voicebridge__voice_models`. Present separate TTS and STT single-choice selectors using Codex's structured question UI. The structured UI starts on the first listed option, so list `current.tts` and `current.stt` first with "(Current)" appended to their labels — or the matching first-run `defaults` entry when an existing custom model is not in the catalog — then the remaining options lightest to heaviest, appending "(Recommended)" to the `defaults` entry when it is not the current one. Show every option's tier, download size, and short description. 4. If the user cancels either selector, end without changing settings. Otherwise call `mcp__voicebridge__voice_configure` with both chosen IDs. 5. Confirm the selected labels and say the models download when the next `$voice-code` session starts. Do not call `mcp__voicebridge__voice_start`, `mcp__voice