Perplexity BYOK TTS — Speak Extension

Free, open-source, Bring-Your-Own-Key text-to-speech for Perplexity AI.

An MIT-licensed Chrome Extension (Manifest V3) that injects a native-looking Play button into Perplexity AI's answer action bar (alongside Copy/Share). Bring your own API key from any TTS provider — OpenAI-compatible or not — and have answers read aloud.

• Zero cost to run — no backend, no telemetry, no monetization. Your keys stay in chrome.storage.
• Pluggable adapters — 12 built-in providers (OpenAI, Groq, Gemini, ElevenLabs, Azure, Polly, Deepgram, Cartesia, Hume, Rime, Unreal Speech, Inworld) + unlimited custom OpenAI-compatible endpoints.
• Auditable build — Chrome Web Store zip is byte-identical to the GitHub repo. What you see is what you get.

Installation (Developer Mode)

1. Clone: git clone https://github.com/your-org/perplexity-byok-tts.git
2. npm install (zero paid deps; build is optional for dev)
3. Open chrome://extensions → enable Developer mode → Load unpacked → select the /extension folder.

How It Works

• content/perplexity.js uses a MutationObserver to find Perplexity's answer action row and injects a Play/Pause/Stop button that inherits Tailwind styles.
• Keys & settings live in chrome.storage.local by default, with an opt-in chrome.storage.sync toggle.
• Playback is routed through background/service-worker.js so content scripts never hold keys directly; audio is queued via lib/audio-queue.js with streaming or full-answer modes.

Project Structure

/extension
  manifest.json
  /background/service-worker.js
  /content/perplexity.js, inject-styles.css
  /adapters/*.js (12 providers + custom + base-adapter)
  /lib/storage.js, chunker.js, audio-queue.js, highlighter.js
  /options/options.html, options.js
  /popup/popup.html, popup.js
  /icons/play.svg, pause.svg, stop.svg, icon*.png

Providers

Provider	Auth	Timestamps
OpenAI	Bearer	—
Groq	Bearer	—
Gemini (Google)	API key query param	—
ElevenLabs	xi-api-key	✅ Forced Alignment
Azure Neural TTS	Ocp-Apim-Subscription-Key	—
Amazon Polly	AWS SigV4	—
Deepgram Aura	Token	—
Cartesia	X-API-Key	—
Hume Octave	API key	—
Rime	Bearer	—
Unreal Speech	Bearer	—
Inworld TTS-2	API key + WebSocket option	—
Custom	User-defined	—

Settings

Options page (chrome://extensions → Details → Extension options) lets you:

• Add/edit/delete API keys per provider
• Pick voice & preview it
• Toggle autoplay, playback mode (full vs. sentence streaming), highlight mode (off / timestamp / estimated), and sync-across-devices

Popup gives quick provider/voice switching and a master on/off toggle.

Privacy

All synthesis requests go directly from your browser to the provider you selected. No intermediate server. Keys are stored locally (or optionally synced via your Google account). No analytics.

License

MIT — see LICENSE.