discoclaw 1.2.4 → 2.0.0

package/.context/voice.md

@@ -29,10 +29,16 @@ Two native npm packages power the Discord voice integration:
 | `src/voice/transcript-mirror.ts` | Posts user transcriptions and bot responses to a text channel |
 | `src/voice/voice-action-flags.ts` | Restricted action subset for voice invocations (messaging + tasks + memory only) |
 | `src/voice/conversation-buffer.ts` | Per-guild conversation ring buffer (10 turns) — stores user/model exchanges in memory; backfills from voice-log channel on join |
+| `src/voice/providers/gemini-live-types.ts` | TypeScript interfaces for Gemini Live: `GeminiLiveOpts`, `GeminiLiveEvent`, `GeminiLiveState` |
+| `src/voice/providers/gemini-live-provider.ts` | Bidirectional WebSocket session wrapper for the Gemini Multimodal Live API — connect/disconnect, audio send/receive, reconnect with exponential backoff |
+| `src/voice/providers/gemini-live-responder.ts` | Bridges `GeminiLiveProvider` audio/text events to Discord `AudioPlayer` playback and `TranscriptMirror` logging |
+| `src/voice/providers/index.ts` | Barrel re-export for Gemini Live provider modules |
 | `src/discord/actions-voice.ts` | Discord action types: `voiceJoin`, `voiceLeave`, `voiceStatus`, `voiceMute`, `voiceDeafen` |
 
 ## Audio Data Flow
 
+### Default pipeline (`voiceProvider: 'pipeline'`)
+
 ```
 User speaks in Discord voice channel
   → @discordjs/voice receiver emits Opus packets per user
@@ -47,6 +53,23 @@ User speaks in Discord voice channel
                     → AudioPlayer → Discord voice connection
 ```
 
+### Gemini Live (`voiceProvider: 'gemini-live'`)
+
+Bypasses separate STT/TTS/AI stages — Gemini handles speech recognition, reasoning, and speech synthesis in a single bidirectional WebSocket session.
+
+```
+User speaks in Discord voice channel
+  → @discordjs/voice receiver emits Opus packets per user
+    → AudioReceiver: allowlist gate → OpusDecoder (48 kHz stereo PCM)
+      → downsample to 16 kHz mono
+        → SttProvider shim → GeminiLiveProvider.sendAudio() (WebSocket)
+          → Gemini Live: STT + reasoning + TTS (server-side)
+            ← audio events (24 kHz mono PCM) + text events
+              → GeminiLiveResponder: upsampleToDiscord (48 kHz stereo)
+                → AudioPlayer → Discord voice connection
+              → onBotResponse callback → TranscriptMirror (text channel)
+```
+
 ## Key Patterns
 
 - **Allowlist gating** — `AudioReceiver` only subscribes to users in `DISCORD_ALLOW_USER_IDS`. Empty allowlist = ignore everyone (fail-closed).
@@ -56,6 +79,8 @@ User speaks in Discord voice channel
 - **Generation-based cancellation** — `VoiceResponder` increments a generation counter on each new transcription. If a newer transcription arrives mid-pipeline, the older one is silently abandoned.
 - **Barge-in** — Gated on a non-empty STT transcription result, not the raw VAD `speaking.start` event. Echo from the bot's own TTS leaking through the user's mic produces empty transcriptions and is ignored. Only when `VoiceResponder.handleTranscription()` receives a non-empty transcript while the player is active does it stop playback and advance the generation counter. This eliminates false positives from echo without relying on a static grace-period timeout.
 - **Conversation ring buffer** — `ConversationBuffer` maintains a per-guild 10-turn ring buffer of user/model exchanges that gets injected into the voice prompt as formatted conversation history. Turns are appended live during a session. On voice join, the buffer backfills from recent voice-log channel messages so context carries across disconnects. The buffer is cleared when the bot leaves the voice channel.
+- **`SttProvider` shim for Gemini Live** — In `gemini-live` mode, the pipeline still uses `AudioReceiver` for Opus decode and downsampling, but replaces the real STT provider with a lightweight shim object that implements the `SttProvider` interface. The shim's `feedAudio()` forwards PCM frames directly to `GeminiLiveProvider.sendAudio()`, while its `start()`/`stop()`/`onTranscription()` are no-ops. This reuses the existing audio-receive path without duplicating Opus decode or downsample logic.
+- **Session rotation timer** — `GeminiLiveProvider` starts a timer on each successful connection that fires at `DISCOCLAW_GEMINI_SESSION_ROTATION_MS` (default 13 min), proactively triggering a graceful reconnect before Gemini's ~15 min server-side session limit. The timer reuses the existing reconnect-with-resume-handle path (ws-039), so audio gap is minimal. The timer is cleared on disconnect and reset on each reconnect. Set to `0` to disable rotation (the server will eventually kill the session).
 - **Re-entrancy guard** — `AudioPipelineManager.startPipeline` uses a `starting` set because `VoiceConnection.subscribe()` synchronously fires a Ready state change.
 - **Error containment** — `VoiceConnectionManager` catches connection errors and destroys the connection to prevent process crashes (e.g. DAVE handshake failures).
 - **Deepgram TTS 2000-char limit** — Deepgram Aura REST TTS returns HTTP 413 (silent failure) for inputs exceeding ~2000 characters. `tts-deepgram.ts` truncates the input to 2000 chars before sending to prevent silent audio dropouts. If the AI response is unexpectedly long (e.g. from a missing `VOICE_STYLE_INSTRUCTION`), the user will still hear a truncated response rather than silence.
@@ -78,8 +103,9 @@ When `voiceEnabled=true`, the post-connect block in `src/index.ts` initializes t
 | `DISCOCLAW_VOICE_ENABLED` | `0` | Master switch |
 | `DISCOCLAW_DISCORD_ACTIONS_VOICE` | `0` | Enable voice action types |
 | `DISCOCLAW_VOICE_AUTO_JOIN` | `0` | Auto-join when allowlisted user enters |
-| `DISCOCLAW_STT_PROVIDER` | `deepgram` | STT backend |
-| `DISCOCLAW_TTS_PROVIDER` | `cartesia` | TTS backend (`cartesia`, `deepgram`, `openai`, `kokoro`) |
+| `DISCOCLAW_VOICE_PIPELINE_PROVIDER` | `pipeline` | Voice pipeline mode: `pipeline` (separate STT/AI/TTS stages) or `gemini-live` (single bidirectional Gemini WebSocket). Requires `GEMINI_API_KEY` when set to `gemini-live`. |
+| `DISCOCLAW_STT_PROVIDER` | `deepgram` | STT backend (used in `pipeline` mode only; ignored in `gemini-live` mode) |
+| `DISCOCLAW_TTS_PROVIDER` | `cartesia` | TTS backend (`cartesia`, `deepgram`, `openai`, `kokoro`) (used in `pipeline` mode only; ignored in `gemini-live` mode) |
 | `DISCOCLAW_VOICE_HOME_CHANNEL` | — | Voice audio channel name/ID used for prompt context (not transcript mirroring) |
 | `DISCOCLAW_VOICE_LOG_CHANNEL` | — | Text channel name/ID where `TranscriptMirror` posts user transcriptions and bot responses; falls back to bootstrap-provided `voiceLogChannelId` if unset |
 | `DISCOCLAW_VOICE_MODEL` | `capable` | AI model tier for voice responses |
@@ -89,5 +115,7 @@ When `voiceEnabled=true`, the post-connect block in `src/index.ts` initializes t
 | `DEEPGRAM_TTS_VOICE` | `aura-2-asteria-en` | Deepgram TTS voice name |
 | `DEEPGRAM_TTS_SPEED` | `1.3` | Deepgram TTS playback speed (range 0.5–1.5) |
 | `CARTESIA_API_KEY` | — | Required for cartesia TTS |
+| `DISCOCLAW_GEMINI_SESSION_ROTATION_MS` | `780000` (13 min) | Time before proactive session rotation in `gemini-live` mode. Must be less than Gemini's ~15 min server-side limit. Set to `0` to disable. |
+| `GEMINI_API_KEY` | — | Required when `DISCOCLAW_VOICE_PIPELINE_PROVIDER=gemini-live`. Authenticates the Gemini Multimodal Live WebSocket session. Also used by the `gemini-api` runtime adapter (see `runtime.md`). |
 | `ANTHROPIC_API_KEY` | — | Enables the Anthropic REST adapter; when set and voice is enabled, voice auto-wires to the direct Messages API path (zero CLI cold-start). See `runtime.md § Anthropic REST Runtime`. |
 | *(built-in)* | — | Telegraphic style instruction hardcoded into every voice AI invocation — front-loads the answer, strips preambles/markdown/filler, keeps responses short for TTS latency. Not an env var; not overridable by `DISCOCLAW_VOICE_SYSTEM_PROMPT`. |

package/.env.example

@@ -90,7 +90,7 @@ DISCORD_GUILD_ID=
 # - PRIMARY_RUNTIME sets the default adapter the instance boots with.
 # - `!models set chat <runtime>` can live-switch the main runtime in memory, but chat runtime swaps do not persist.
 # - Persistent model-role defaults live in data/models.json.
-# - Persistent runtime-only overlays live in data/runtime-overrides.json (`voiceRuntime`, `fastRuntime`, `ttsVoice`).
+# - Persistent runtime-only overlays live in data/runtime-overrides.json (`voiceRuntime`, `fastRuntime`).
 # - `!models reset` writes startup-default model strings back into models.json and clears fast/voice runtime overlays;
 #   it does not rewrite PRIMARY_RUNTIME in .env.
 # Supported runtime-path notes:
@@ -191,8 +191,12 @@ DISCORD_GUILD_ID=
 # Voice — configure via `pnpm setup` or `discoclaw init`
 # ----------------------------------------------------------
 # Run `pnpm setup` or `discoclaw init` to enable voice interactively,
-# or set these vars manually to enable voice chat (STT/TTS via Deepgram).
+# or set these vars manually to enable Gemini Live voice chat.
 #DISCOCLAW_VOICE_ENABLED=0
+# Gemini Live requires GEMINI_API_KEY.
+# Gemini Live session rotation threshold (ms). The provider proactively reconnects
+# before Gemini's ~15 min session limit to minimize audio gap. Default: 780000 (13 min).
+#DISCOCLAW_GEMINI_SESSION_ROTATION_MS=780000
 # Text channel used for voice prompt context and actions (e.g. posting action results,
 # reading pinned notes). Required for full voice functionality when voice is enabled.
 #DISCOCLAW_VOICE_HOME_CHANNEL= # e.g. "voice"
@@ -200,7 +204,7 @@ DISCORD_GUILD_ID=
 # (the bot creates a "voice-log" text channel and stores its ID in system-scaffold.json).
 # Only set this to override the auto-discovered channel.
 #DISCOCLAW_VOICE_LOG_CHANNEL=
-#DEEPGRAM_API_KEY=
+#GEMINI_API_KEY=
 # Optional voice-only Anthropic runtime for voice responses.
 # `claude-api` is not a valid PRIMARY_RUNTIME and does not persist in models.json.
 # Use `!models set voice claude-api` to persist the voice runtime path in runtime-overrides.json.

package/.env.example.full

@@ -703,32 +703,22 @@ DISCOCLAW_DISCORD_ACTIONS_IMAGEGEN=0
 # IMAGEGEN_GEMINI_API_KEY is set.
 
 # ----------------------------------------------------------
-# Voice (STT/TTS) — join voice channels, listen and respond
+# Voice — Gemini Live voice chat
 # ----------------------------------------------------------
-# Master switch — enables voice channel interaction (default: off).
-# When enabled, the bot can join Discord voice channels, transcribe speech via STT,
-# and respond with synthesized speech via TTS.
+# Master switch — enables Discord voice interaction (default: off).
+# Gemini Live handles speech recognition, reasoning, and speech synthesis in one session.
 #DISCOCLAW_VOICE_ENABLED=0
 # Enable voice Discord action category (voiceJoin, voiceLeave, voiceStatus, voiceMute, voiceDeafen).
 # Requires DISCOCLAW_VOICE_ENABLED=1 to take effect (default: off).
 #DISCOCLAW_DISCORD_ACTIONS_VOICE=0
 # Auto-join voice channels when a non-bot user joins, and auto-leave when the last
-# non-bot user leaves. Starts/tears down the audio pipeline (STT receiver) automatically.
-# Requires DISCOCLAW_VOICE_ENABLED=1 (default: off).
+# non-bot user leaves. Requires DISCOCLAW_VOICE_ENABLED=1 (default: off).
 #DISCOCLAW_VOICE_AUTO_JOIN=0
-# Speech-to-text provider: deepgram (Deepgram Nova-3 API) or whisper (whisper.cpp local).
-# deepgram requires DEEPGRAM_API_KEY; whisper runs locally with no API key.
-#DISCOCLAW_STT_PROVIDER=deepgram
-# Text-to-speech provider: cartesia | deepgram | kokoro | openai.
-# cartesia requires CARTESIA_API_KEY; deepgram reuses DEEPGRAM_API_KEY;
-# openai requires OPENAI_API_KEY; kokoro runs locally with no API key.
-#DISCOCLAW_TTS_PROVIDER=cartesia
-# Voice audio channel name or ID — the channel the bot joins for voice interaction.
-# Used as the prompt context source (root policy, PA files, channel context, durable memory).
+# Voice text channel name or ID — used for prompt context and voice-triggered actions.
 # The old env var DISCOCLAW_VOICE_TRANSCRIPT_CHANNEL is still accepted as a fallback.
 # Leave unset to skip voice channel context in prompts.
 #DISCOCLAW_VOICE_HOME_CHANNEL=  # e.g. "voice" if using the default scaffold
-# Text channel name or ID for posting voice transcripts (user STT and bot TTS responses).
+# Text channel name or ID for posting voice transcripts.
 # Optional — auto-discovered via bootstrap (the bot creates "voice-log" and stores its ID
 # in system-scaffold.json). Only set this to override the auto-discovered channel.
 # Leave unset to disable transcript mirroring.
@@ -741,20 +731,11 @@ DISCOCLAW_DISCORD_ACTIONS_IMAGEGEN=0
 # Custom system prompt prepended to voice AI invocations. Max 4000 chars.
 # Use this to set a conversational tone, brevity instructions, or persona for voice responses.
 #DISCOCLAW_VOICE_SYSTEM_PROMPT=
-# Anthropic API key for direct Messages API access (bypasses Claude CLI cold-start).
-# When set and voice is enabled, voice invocations use the Anthropic REST adapter
-# instead of the CLI subprocess, eliminating ~2-5s cold-start latency per response.
+# Gemini Live session rotation threshold (ms). The provider proactively reconnects
+# before Gemini's ~15 min session limit to minimize audio gap. Default: 780000 (13 min).
+#DISCOCLAW_GEMINI_SESSION_ROTATION_MS=780000
+# Google Gemini API key. Required when voice is enabled.
+#GEMINI_API_KEY=
+# Anthropic API key for the optional direct Messages API voice runtime.
+# When set, `!models set voice claude-api` can bypass CLI cold-start for voice responses.
 #ANTHROPIC_API_KEY=
-# API key for Deepgram Nova-3 STT. Required when DISCOCLAW_STT_PROVIDER=deepgram.
-#DEEPGRAM_API_KEY=
-# Deepgram STT model for voice transcription (default: nova-3-conversationalai).
-# See https://developers.deepgram.com/docs/models-languages-overview for available models.
-#DEEPGRAM_STT_MODEL=nova-3-conversationalai
-# Deepgram TTS voice for speech synthesis (default: aura-2-asteria-en).
-# See https://developers.deepgram.com/docs/tts-models for available voices.
-#DEEPGRAM_TTS_VOICE=aura-2-asteria-en
-# Deepgram TTS playback speed (range: 0.5–1.5, default: 1.3).
-# Values below 1.0 slow down speech; values above 1.0 speed it up.
-#DEEPGRAM_TTS_SPEED=1.3
-# API key for Cartesia Sonic-3 TTS. Required when DISCOCLAW_TTS_PROVIDER=cartesia.
-#CARTESIA_API_KEY=

package/README.md

@@ -39,7 +39,7 @@ No gateways, no proxies, no web UI. Discord *is* the interface.
 
 ## Voice — the bot talks back
 
-Real-time voice with STT (Deepgram), TTS (Cartesia), barge-in, and transcript mirroring. Off by default. [Setup guide →](docs/voice.md)
+Real-time voice with Gemini Live, barge-in, tool calls, and transcript mirroring. Off by default. [Setup guide →](docs/voice.md)
 
 ## Self-management
 

package/dist/cli/dashboard.js

@@ -146,7 +146,13 @@ function normalizeRuntimeName(value) {
     const trimmed = value?.trim().toLowerCase();
     if (!trimmed)
         return undefined;
-    const normalized = trimmed === 'claude_code' ? 'claude' : trimmed;
+    let normalized = trimmed === 'claude_code' ? 'claude' : trimmed;
+    if (normalized === 'claude-cli')
+        normalized = 'claude';
+    if (normalized === 'codex-cli')
+        normalized = 'codex';
+    if (normalized === 'claude' || normalized === 'codex')
+        return normalized;
     return KNOWN_RUNTIMES.has(normalized) ? normalized : undefined;
 }
 function trimEnvValue(value) {

package/dist/cli/dashboard.test.js

@@ -591,7 +591,6 @@ describe('runDashboard', () => {
             runtimeOverrides: {
                 fastRuntime: 'openrouter',
                 voiceRuntime: 'anthropic',
-                ttsVoice: 'alloy',
             },
             envDefaults: {
                 ...makeDoctorContext().envDefaults,
@@ -628,7 +627,6 @@ describe('runDashboard', () => {
         expect(saveModelConfigMock).toHaveBeenCalledWith('/repo/data/models.json', {});
         expect(saveOverridesMock).toHaveBeenCalledWith('/repo/data/runtime-overrides.json', {
             voiceRuntime: 'anthropic',
-            ttsVoice: 'alloy',
         });
         expect(frames.some((frame) => frame.includes('Reset fast to default: capable. Cleared fastRuntime override. Changes take effect on next service restart.'))).toBe(true);
     });
@@ -638,7 +636,6 @@ describe('runDashboard', () => {
             runtimeOverrides: {
                 fastRuntime: 'openrouter',
                 voiceRuntime: 'anthropic',
-                ttsVoice: 'alloy',
             },
             envDefaults: {
                 ...makeDoctorContext().envDefaults,
@@ -675,7 +672,6 @@ describe('runDashboard', () => {
         expect(saveModelConfigMock).toHaveBeenCalledWith('/repo/data/models.json', {});
         expect(saveOverridesMock).toHaveBeenCalledWith('/repo/data/runtime-overrides.json', {
             fastRuntime: 'openrouter',
-            ttsVoice: 'alloy',
         });
         expect(frames.some((frame) => frame.includes('Reset voice to default: capable. Cleared voiceRuntime override. Changes take effect on next service restart.'))).toBe(true);
     });

package/dist/cli/init-wizard.js

@@ -88,10 +88,8 @@ export function buildEnvContent(vals, now = new Date()) {
     // Voice
     const voiceKeys = [
         'DISCOCLAW_VOICE_ENABLED',
-        'DEEPGRAM_API_KEY',
+        'GEMINI_API_KEY',
         'DISCOCLAW_DISCORD_ACTIONS_VOICE',
-        'DISCOCLAW_STT_PROVIDER',
-        'DISCOCLAW_TTS_PROVIDER',
     ];
     const hasVoice = voiceKeys.some((k) => vals[k]);
     if (hasVoice) {
@@ -347,14 +345,12 @@ export async function runInitWizard() {
     }
     values.DISCOCLAW_DISCORD_ACTIONS = '1';
     // ── Voice setup ───────────────────────────────────────────────────────────
-    const enableVoice = await ask('\nEnable voice chat? (requires a Deepgram API key — you can skip this and enable later) [y/N] ');
+    const enableVoice = await ask('\nEnable voice chat? (requires a Gemini API key — you can skip this and enable later) [y/N] ');
     if (enableVoice.toLowerCase() === 'y') {
-        const deepgramKey = await askValidated('Deepgram API key: ', (val) => (val ? null : 'Deepgram API key is required'));
+        const geminiKey = await askValidated('Gemini API key: ', (val) => (val ? null : 'Gemini API key is required'));
         values.DISCOCLAW_VOICE_ENABLED = '1';
-        values.DEEPGRAM_API_KEY = deepgramKey;
+        values.GEMINI_API_KEY = geminiKey;
         values.DISCOCLAW_DISCORD_ACTIONS_VOICE = '1';
-        values.DISCOCLAW_STT_PROVIDER = 'deepgram';
-        values.DISCOCLAW_TTS_PROVIDER = 'deepgram';
     }
     // ── Write .env ────────────────────────────────────────────────────────────
     const envContent = buildEnvContent(values);

package/dist/cli/init-wizard.test.js

@@ -155,16 +155,12 @@ describe('init wizard helpers', () => {
             DISCORD_TOKEN: 'a.b.c',
             DISCORD_ALLOW_USER_IDS: '1000000000000000001',
             DISCOCLAW_VOICE_ENABLED: '1',
-            DEEPGRAM_API_KEY: 'dg-key',
+            GEMINI_API_KEY: 'gm-key',
             DISCOCLAW_DISCORD_ACTIONS_VOICE: '1',
-            DISCOCLAW_STT_PROVIDER: 'deepgram',
-            DISCOCLAW_TTS_PROVIDER: 'deepgram',
         }, new Date('2026-02-26T00:00:00.000Z'));
         expect(content).toContain('# VOICE');
         expect(content).toContain('DISCOCLAW_VOICE_ENABLED=1');
-        expect(content).toContain('DEEPGRAM_API_KEY=dg-key');
-        expect(content).toContain('DISCOCLAW_STT_PROVIDER=deepgram');
-        expect(content).toContain('DISCOCLAW_TTS_PROVIDER=deepgram');
+        expect(content).toContain('GEMINI_API_KEY=gm-key');
     });
     it('omits voice section when no voice vars are provided', () => {
         const content = buildEnvContent({
@@ -555,7 +551,7 @@ describe('runInitWizard', () => {
             '5000000000000000001', // DISCORD_GUILD_ID
             '', // provider selection -> default (Claude)
             'y', // enable voice -> yes
-            'dg-test-key', // Deepgram API key
+            'gemini-test-key', // Gemini API key
         ];
         process.chdir(tmpDir);
         vi.mocked(createInterface).mockReturnValue(makeReadline(answers));
@@ -573,9 +569,7 @@ describe('runInitWizard', () => {
         const newEnv = fs.readFileSync(path.join(tmpDir, '.env'), 'utf8');
         expect(newEnv).toContain('# VOICE');
         expect(newEnv).toContain('DISCOCLAW_VOICE_ENABLED=1');
-        expect(newEnv).toContain('DEEPGRAM_API_KEY=dg-test-key');
-        expect(newEnv).toContain('DISCOCLAW_STT_PROVIDER=deepgram');
-        expect(newEnv).toContain('DISCOCLAW_TTS_PROVIDER=deepgram');
+        expect(newEnv).toContain('GEMINI_API_KEY=gemini-test-key');
         expect(newEnv).toContain('DISCOCLAW_DISCORD_ACTIONS_VOICE=1');
     });
 });

package/dist/config.js

@@ -510,8 +510,7 @@ export function parseConfig(env) {
     const anthropicApiKey = parseTrimmedString(env, 'ANTHROPIC_API_KEY');
     const voiceEnabled = parseBoolean(env, 'DISCOCLAW_VOICE_ENABLED', false);
     const voiceAutoJoin = parseBoolean(env, 'DISCOCLAW_VOICE_AUTO_JOIN', false);
-    const voiceSttProvider = parseEnum(env, 'DISCOCLAW_STT_PROVIDER', ['deepgram', 'whisper', 'openai'], 'deepgram');
-    const voiceTtsProvider = parseEnum(env, 'DISCOCLAW_TTS_PROVIDER', ['cartesia', 'deepgram', 'kokoro', 'openai'], 'cartesia');
+    const geminiSessionRotationMs = parseNonNegativeInt(env, 'DISCOCLAW_GEMINI_SESSION_ROTATION_MS', 780_000);
     let voiceHomeChannel = parseTrimmedString(env, 'DISCOCLAW_VOICE_HOME_CHANNEL');
     if (!voiceHomeChannel) {
         const legacy = parseTrimmedString(env, 'DISCOCLAW_VOICE_TRANSCRIPT_CHANNEL');
@@ -521,20 +520,6 @@ export function parseConfig(env) {
         }
     }
     const voiceLogChannel = parseTrimmedString(env, 'DISCOCLAW_VOICE_LOG_CHANNEL');
-    const deepgramApiKey = parseTrimmedString(env, 'DEEPGRAM_API_KEY');
-    const deepgramSttModel = parseTrimmedString(env, 'DEEPGRAM_STT_MODEL') ?? 'nova-3-general';
-    const deepgramTtsVoice = parseTrimmedString(env, 'DEEPGRAM_TTS_VOICE') ?? 'aura-2-asteria-en';
-    const deepgramTtsSpeed = (() => {
-        const raw = parseTrimmedString(env, 'DEEPGRAM_TTS_SPEED');
-        if (raw == null)
-            return 1.3;
-        const n = parseFloat(raw);
-        if (!Number.isFinite(n) || n < 0.5 || n > 1.5) {
-            throw new Error(`DEEPGRAM_TTS_SPEED must be a number between 0.5 and 1.5, got "${raw}"`);
-        }
-        return n;
-    })();
-    const cartesiaApiKey = parseTrimmedString(env, 'CARTESIA_API_KEY');
     const voiceModelRaw = parseTrimmedString(env, 'DISCOCLAW_VOICE_MODEL');
     const voiceSystemPrompt = (() => {
         const raw = parseTrimmedString(env, 'DISCOCLAW_VOICE_SYSTEM_PROMPT');
@@ -545,24 +530,12 @@ export function parseConfig(env) {
         }
         return raw;
     })();
-    if (voiceEnabled && voiceSttProvider === 'deepgram' && !deepgramApiKey) {
-        warnings.push('DISCOCLAW_VOICE_ENABLED=1 with STT provider "deepgram" but DEEPGRAM_API_KEY is not set; voice STT will fail at runtime.');
-    }
-    if (voiceEnabled && voiceSttProvider === 'openai' && !openaiApiKey) {
-        warnings.push('DISCOCLAW_VOICE_ENABLED=1 with STT provider "openai" but OPENAI_API_KEY is not set; voice STT will fail at runtime.');
-    }
-    if (voiceEnabled && voiceTtsProvider === 'cartesia' && !cartesiaApiKey) {
-        warnings.push('DISCOCLAW_VOICE_ENABLED=1 with TTS provider "cartesia" but CARTESIA_API_KEY is not set; voice TTS will fail at runtime.');
-    }
-    if (voiceEnabled && voiceTtsProvider === 'deepgram' && !deepgramApiKey) {
-        warnings.push('DISCOCLAW_VOICE_ENABLED=1 with TTS provider "deepgram" but DEEPGRAM_API_KEY is not set; voice TTS will fail at runtime.');
-    }
-    if (voiceEnabled && voiceTtsProvider === 'openai' && !openaiApiKey) {
-        warnings.push('DISCOCLAW_VOICE_ENABLED=1 with TTS provider "openai" but OPENAI_API_KEY is not set; voice TTS will fail at runtime.');
-    }
     if (voiceEnabled && !voiceHomeChannel) {
         warnings.push('DISCOCLAW_VOICE_ENABLED=1 but DISCOCLAW_VOICE_HOME_CHANNEL is not set; voice actions will be disabled (no target channel for action execution).');
     }
+    if (voiceEnabled && !geminiApiKey) {
+        warnings.push('DISCOCLAW_VOICE_ENABLED=1 but GEMINI_API_KEY is not set; Gemini Live voice will fail at runtime.');
+    }
     const coldStorageEnabled = parseBoolean(env, 'DISCOCLAW_COLD_STORAGE_ENABLED', false);
     const coldStorageApiKey = parseTrimmedString(env, 'COLD_STORAGE_API_KEY') ?? openaiApiKey;
     const coldStorageProvider = parseEnum(env, 'COLD_STORAGE_PROVIDER', ['openai', 'openai-compat'], 'openai');
@@ -741,15 +714,9 @@ export function parseConfig(env) {
             voiceAutoJoin,
             voiceModel,
             voiceSystemPrompt,
-            voiceSttProvider,
-            voiceTtsProvider,
+            geminiSessionRotationMs,
             voiceHomeChannel,
             voiceLogChannel,
-            deepgramApiKey,
-            deepgramSttModel,
-            deepgramTtsVoice,
-            deepgramTtsSpeed,
-            cartesiaApiKey,
             forgeDrafterRuntime,
             forgeAuditorRuntime,
             openrouterApiKey,

package/dist/config.test.js

@@ -1021,41 +1021,9 @@ describe('parseConfig', () => {
         expect(config.voiceEnabled).toBe(false);
     });
     it('parses DISCOCLAW_VOICE_ENABLED=1 as true', () => {
-        const { config } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', DEEPGRAM_API_KEY: 'dg-key', CARTESIA_API_KEY: 'ca-key' }));
+        const { config } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', GEMINI_API_KEY: 'gm-key' }));
         expect(config.voiceEnabled).toBe(true);
     });
-    it('defaults voiceSttProvider to "deepgram"', () => {
-        const { config } = parseConfig(env());
-        expect(config.voiceSttProvider).toBe('deepgram');
-    });
-    it('parses DISCOCLAW_STT_PROVIDER=whisper', () => {
-        const { config } = parseConfig(env({ DISCOCLAW_STT_PROVIDER: 'whisper' }));
-        expect(config.voiceSttProvider).toBe('whisper');
-    });
-    it('parses STT provider case-insensitively', () => {
-        const { config } = parseConfig(env({ DISCOCLAW_STT_PROVIDER: 'Deepgram' }));
-        expect(config.voiceSttProvider).toBe('deepgram');
-    });
-    it('throws on invalid STT provider', () => {
-        expect(() => parseConfig(env({ DISCOCLAW_STT_PROVIDER: 'invalid' })))
-            .toThrow(/DISCOCLAW_STT_PROVIDER must be one of deepgram\|whisper/);
-    });
-    it('defaults voiceTtsProvider to "cartesia"', () => {
-        const { config } = parseConfig(env());
-        expect(config.voiceTtsProvider).toBe('cartesia');
-    });
-    it('parses DISCOCLAW_TTS_PROVIDER=kokoro', () => {
-        const { config } = parseConfig(env({ DISCOCLAW_TTS_PROVIDER: 'kokoro' }));
-        expect(config.voiceTtsProvider).toBe('kokoro');
-    });
-    it('parses TTS provider case-insensitively', () => {
-        const { config } = parseConfig(env({ DISCOCLAW_TTS_PROVIDER: 'Cartesia' }));
-        expect(config.voiceTtsProvider).toBe('cartesia');
-    });
-    it('throws on invalid TTS provider', () => {
-        expect(() => parseConfig(env({ DISCOCLAW_TTS_PROVIDER: 'elevenlabs' })))
-            .toThrow(/DISCOCLAW_TTS_PROVIDER must be one of cartesia\|deepgram\|kokoro\|openai/);
-    });
     it('parses DISCOCLAW_VOICE_HOME_CHANNEL when set', () => {
         const { config } = parseConfig(env({ DISCOCLAW_VOICE_HOME_CHANNEL: 'voice-log' }));
         expect(config.voiceHomeChannel).toBe('voice-log');
@@ -1085,33 +1053,13 @@ describe('parseConfig', () => {
         const { config } = parseConfig(env());
         expect(config.voiceLogChannel).toBeUndefined();
     });
-    it('parses DEEPGRAM_API_KEY when set', () => {
-        const { config } = parseConfig(env({ DEEPGRAM_API_KEY: 'dg-key' }));
-        expect(config.deepgramApiKey).toBe('dg-key');
+    it('warns when voice enabled but GEMINI_API_KEY is missing', () => {
+        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1' }));
+        expect(warnings.some((w) => w.includes('GEMINI_API_KEY'))).toBe(true);
     });
-    it('returns undefined for deepgramApiKey when unset', () => {
-        const { config } = parseConfig(env());
-        expect(config.deepgramApiKey).toBeUndefined();
-    });
-    it('parses CARTESIA_API_KEY when set', () => {
-        const { config } = parseConfig(env({ CARTESIA_API_KEY: 'ca-key' }));
-        expect(config.cartesiaApiKey).toBe('ca-key');
-    });
-    it('returns undefined for cartesiaApiKey when unset', () => {
-        const { config } = parseConfig(env());
-        expect(config.cartesiaApiKey).toBeUndefined();
-    });
-    it('warns when voice enabled with deepgram STT but DEEPGRAM_API_KEY missing', () => {
-        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', CARTESIA_API_KEY: 'ca-key' }));
-        expect(warnings.some((w) => w.includes('DEEPGRAM_API_KEY'))).toBe(true);
-    });
-    it('does not warn about DEEPGRAM_API_KEY when voice disabled', () => {
+    it('does not warn about GEMINI_API_KEY when voice disabled', () => {
         const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '0' }));
-        expect(warnings.some((w) => w.includes('DEEPGRAM_API_KEY'))).toBe(false);
-    });
-    it('does not warn about DEEPGRAM_API_KEY when STT provider is whisper', () => {
-        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', DISCOCLAW_STT_PROVIDER: 'whisper', CARTESIA_API_KEY: 'ca-key' }));
-        expect(warnings.some((w) => w.includes('DEEPGRAM_API_KEY'))).toBe(false);
+        expect(warnings.some((w) => w.includes('GEMINI_API_KEY'))).toBe(false);
     });
     // --- voiceAutoJoin ---
     it('defaults voiceAutoJoin to false', () => {
@@ -1130,24 +1078,12 @@ describe('parseConfig', () => {
         const { config } = parseConfig(env({ DISCOCLAW_VOICE_AUTO_JOIN: '0' }));
         expect(config.voiceAutoJoin).toBe(false);
     });
-    it('warns when voice enabled with cartesia TTS but CARTESIA_API_KEY missing', () => {
-        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', DEEPGRAM_API_KEY: 'dg-key' }));
-        expect(warnings.some((w) => w.includes('CARTESIA_API_KEY'))).toBe(true);
-    });
-    it('does not warn about CARTESIA_API_KEY when voice disabled', () => {
-        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '0' }));
-        expect(warnings.some((w) => w.includes('CARTESIA_API_KEY'))).toBe(false);
-    });
-    it('does not warn about CARTESIA_API_KEY when TTS provider is kokoro', () => {
-        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', DISCOCLAW_TTS_PROVIDER: 'kokoro', DEEPGRAM_API_KEY: 'dg-key' }));
-        expect(warnings.some((w) => w.includes('CARTESIA_API_KEY'))).toBe(false);
-    });
     it('warns when voice enabled but DISCOCLAW_VOICE_HOME_CHANNEL is unset', () => {
-        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', DEEPGRAM_API_KEY: 'dg-key', CARTESIA_API_KEY: 'ca-key' }));
+        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', GEMINI_API_KEY: 'gm-key' }));
         expect(warnings.some((w) => w.includes('DISCOCLAW_VOICE_HOME_CHANNEL'))).toBe(true);
     });
     it('does not warn about DISCOCLAW_VOICE_HOME_CHANNEL when both voice and home channel are set', () => {
-        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', DISCOCLAW_VOICE_HOME_CHANNEL: '1000000000000000003', DEEPGRAM_API_KEY: 'dg-key', CARTESIA_API_KEY: 'ca-key' }));
+        const { warnings } = parseConfig(env({ DISCOCLAW_VOICE_ENABLED: '1', DISCOCLAW_VOICE_HOME_CHANNEL: '1000000000000000003', GEMINI_API_KEY: 'gm-key' }));
         expect(warnings.some((w) => w.includes('DISCOCLAW_VOICE_HOME_CHANNEL'))).toBe(false);
     });
     // --- cold storage ---