🤖 feat: voice input mode with OpenAI transcription

bun.lock

@@ -42,6 +42,7 @@
         "minimist": "^1.2.8",
         "motion": "^12.23.24",
         "ollama-ai-provider-v2": "^1.5.4",
+        "openai": "^6.9.1",
         "rehype-harden": "^1.1.5",
         "shescape": "^2.1.6",
         "source-map-support": "^0.5.21",
@@ -2688,6 +2689,8 @@
 
     "oniguruma-to-es": ["oniguruma-to-es@4.3.3", "", { "dependencies": { "oniguruma-parser": "^0.12.1", "regex": "^6.0.1", "regex-recursion": "^6.0.2" } }, "sha512-rPiZhzC3wXwE59YQMRDodUwwT9FZ9nNBwQQfsd1wfdtlKEyCdRV0avrTcSZ5xlIvGRVPd/cx6ZN45ECmS39xvg=="],
 
+    "openai": ["openai@6.9.1", "", { "peerDependencies": { "ws": "^8.18.0", "zod": "^3.25 || ^4.0" }, "optionalPeers": ["ws", "zod"], "bin": { "openai": "bin/cli" } }, "sha512-vQ5Rlt0ZgB3/BNmTa7bIijYFhz3YBceAA3Z4JuoMSBftBF9YqFHIEhZakSs+O/Ad7EaoEimZvHxD5ylRjN11Lg=="],
+
     "optionator": ["optionator@0.9.4", "", { "dependencies": { "deep-is": "^0.1.3", "fast-levenshtein": "^2.0.6", "levn": "^0.4.1", "prelude-ls": "^1.2.1", "type-check": "^0.4.0", "word-wrap": "^1.2.5" } }, "sha512-6IpQ7mKUxRcZNLIObR0hz7lxsapSSIYNZJwXPGeF0mTVqGKFIXj1DQcMoT22S3ROcLyY/rz0PWaWZ9ayWmad9g=="],
 
     "ora": ["ora@5.4.1", "", { "dependencies": { "bl": "^4.1.0", "chalk": "^4.1.0", "cli-cursor": "^3.1.0", "cli-spinners": "^2.5.0", "is-interactive": "^1.0.0", "is-unicode-supported": "^0.1.0", "log-symbols": "^4.1.0", "strip-ansi": "^6.0.0", "wcwidth": "^1.0.1" } }, "sha512-5b6Y85tPxZZ7QytO+BQzysW31HJku27cRIlkbAXaNx+BdcVi+LlRFmVXzeF6a7JCwJpyw5c4b+YSVImQIrBpuQ=="],

package.json

@@ -83,6 +83,7 @@
     "minimist": "^1.2.8",
     "motion": "^12.23.24",
     "ollama-ai-provider-v2": "^1.5.4",
+    "openai": "^6.9.1",
     "rehype-harden": "^1.1.5",
     "shescape": "^2.1.6",
     "source-map-support": "^0.5.21",

src/browser/api.ts

@@ -361,6 +361,9 @@ const webApi: IPCApi = {
     },
     closeWindow: (workspaceId) => invokeIPC(IPC_CHANNELS.TERMINAL_WINDOW_CLOSE, workspaceId),
   },
+  voice: {
+    transcribe: (audioBase64) => invokeIPC(IPC_CHANNELS.VOICE_TRANSCRIBE, audioBase64),
+  },
   update: {
     check: () => invokeIPC(IPC_CHANNELS.UPDATE_CHECK),
     download: () => invokeIPC(IPC_CHANNELS.UPDATE_DOWNLOAD),

src/browser/components/ChatInput/VoiceInputButton.tsx

@@ -0,0 +1,74 @@
+/**
+ * Voice input button - floats inside the chat input textarea.
+ * Minimal footprint: just an icon that changes color based on state.
+ */
+
+import React from "react";
+import { Mic, Loader2 } from "lucide-react";
+import { TooltipWrapper, Tooltip } from "../Tooltip";
+import { formatKeybind, KEYBINDS } from "@/browser/utils/ui/keybinds";
+import { cn } from "@/common/lib/utils";
+import type { VoiceInputState } from "@/browser/hooks/useVoiceInput";
+
+interface VoiceInputButtonProps {
+  state: VoiceInputState;
+  isApiKeySet: boolean;
+  shouldShowUI: boolean;
+  onToggle: () => void;
+  disabled?: boolean;
+}
+
+const STATE_CONFIG: Record<VoiceInputState, { label: string; colorClass: string }> = {
+  idle: { label: "Voice input", colorClass: "text-muted/50 hover:text-muted" },
+  recording: { label: "Stop recording", colorClass: "text-blue-500 animate-pulse" },
+  transcribing: { label: "Transcribing...", colorClass: "text-amber-500" },
+};
+
+export const VoiceInputButton: React.FC<VoiceInputButtonProps> = (props) => {
+  if (!props.shouldShowUI) return null;
+
+  const needsApiKey = !props.isApiKeySet;
+  const { label, colorClass } = needsApiKey
+    ? { label: "Voice input (requires OpenAI API key)", colorClass: "text-muted/50" }
+    : STATE_CONFIG[props.state];
+
+  const Icon = props.state === "transcribing" ? Loader2 : Mic;
+  const isTranscribing = props.state === "transcribing";
+
+  return (
+    <TooltipWrapper inline>
+      <button
+        type="button"
+        onClick={props.onToggle}
+        disabled={(props.disabled ?? false) || isTranscribing || needsApiKey}
+        aria-label={label}
+        aria-pressed={props.state === "recording"}
+        className={cn(
+          "inline-flex items-center justify-center rounded p-0.5 transition-colors duration-150",
+          "disabled:cursor-not-allowed disabled:opacity-40",
+          colorClass
+        )}
+      >
+        <Icon className={cn("h-4 w-4", isTranscribing && "animate-spin")} strokeWidth={1.5} />
+      </button>
+      <Tooltip className="tooltip" align="right">
+        {needsApiKey ? (
+          <>
+            Voice input requires OpenAI API key.
+            <br />
+            Configure in Settings → Providers.
+          </>
+        ) : (
+          <>
+            <strong>Voice input</strong> — press space on empty input
+            <br />
+            or {formatKeybind(KEYBINDS.TOGGLE_VOICE_INPUT)} anytime
+            <br />
+            <br />
+            While recording: space sends, esc cancels
+          </>
+        )}
+      </Tooltip>
+    </TooltipWrapper>
+  );
+};

src/browser/components/ChatInput/WaveformBars.tsx

@@ -0,0 +1,32 @@
+/**
+ * Animated waveform bars for voice recording UI.
+ * Shows 5 bars with staggered pulse animation.
+ */
+
+import { cn } from "@/common/lib/utils";
+
+interface WaveformBarsProps {
+  /** Color class for the bars (e.g., "bg-blue-500") */
+  colorClass: string;
+  /** Whether to mirror the animation (for right-side waveform) */
+  mirrored?: boolean;
+}
+
+export const WaveformBars: React.FC<WaveformBarsProps> = (props) => {
+  const indices = props.mirrored ? [4, 3, 2, 1, 0] : [0, 1, 2, 3, 4];
+
+  return (
+    <div className="flex items-center gap-1">
+      {indices.map((i, displayIndex) => (
+        <div
+          key={displayIndex}
+          className={cn("w-1 rounded-full", props.colorClass)}
+          style={{
+            height: `${12 + Math.sin(i * 0.8) * 8}px`,
+            animation: `pulse 0.8s ease-in-out ${i * 0.1}s infinite alternate`,
+          }}
+        />
+      ))}
+    </div>
+  );
+};