AI-Native Text-to-Speech

One API.Eight Providers.

Bark is a unified TTS gateway that routes your text to the best voice provider — locally, privately, and with full cost visibility. Free and paid providers, one interface.

Explore FeaturesQuick Start

8

Providers

400+

Voices

100+

Languages

Why Bark

Intelligent TTS, your way.

Stop juggling provider SDKs. Bark gives you one unified interface with smart defaults and full control when you need it.

Smart Routing

Language-aware provider selection. Chinese text automatically routes to the best CJK voice; English to native providers. Configurable fallback chains.

Cost Tracking

Automatic per-request cost estimation with a local usage ledger. See exactly how much each provider costs you — no surprises.

Provider Comparison

Synthesize the same text across multiple providers simultaneously. Compare latency, cost, and quality side-by-side.

Local-First

Your config, usage data, and audio files stay on your machine. No cloud account required — just API keys for paid providers.

Provider Network

Eight providers, one gateway.

Mix free and paid providers. Route by language, quality tier, or cost. Bark picks the right voice automatically.

Free

Microsoft Edge TTS

400+ voices, 100+ languages

Free

Google Translate TTS

50+ languages

Paid

OpenAI TTS

HD quality, natural voices

Paid

Azure Speech

500+ neural voices

Paid

Google Cloud TTS

WaveNet & Neural2 voices

Paid

ElevenLabs

Ultra-realistic, voice cloning

Paid

MiniMax

Chinese-optimized

Paid

Volcengine

ByteDance, CJK-focused

Routing Configuration

routing:
  default_provider: "edge_tts"
  rules:
    - match: { language: "zh-*" }
      prefer: [edge_tts, volcengine, minimax, azure]
    - match: { language: "en-*" }
      prefer: [openai, elevenlabs, edge_tts]
    - match: { language: "*" }
      prefer: [edge_tts, google_translate]

Use It Your Way

CLI. Python. HTTP.

Whether you're scripting from the terminal, building a Python app, or integrating via REST — Bark meets you where you are.

CLI
obk speak "Hello world" --provider edge_tts

One command to speak. Pipe text, choose providers, save files.

Python API
from orca_bark import speak_to_file

speak_to_file("Hello", "hello.mp3",
              provider="openai")

Import and call. Async-ready, type-hinted, zero boilerplate.

HTTP API
curl -X POST http://localhost:8723/v1/speak \
  -H "Authorization: Bearer $KEY" \
  -d '{"text":"Hello","provider":"edge_tts"}' \
  --output hello.mp3

REST endpoint with OpenAPI docs. Drop into any stack.

Native macOS Experience

Select text. Press a shortcut.Hear it spoken.

The Bark macOS app lives in your menu bar. Select any text in any app, hit your keyboard shortcut, and it speaks — powered by your configured providers. Built with SwiftUI, runs natively on Apple Silicon.

Menu Bar App

Always accessible from your menu bar. No dock icon, no window clutter.

System-Wide Shortcut

Bind a global hotkey to speak selected text from any application.

Self-Contained

Embedded Python runtime — no Homebrew, no pip install. Just drag to Applications.

Get started in seconds.

# Install with pip
pip install orca-bark

# Or with uv
uv pip install orca-bark

# Speak immediately (free, no API key)
obk speak "Hello from Bark"

# Start the HTTP server
obk serve --port 8723

MIT License. Free providers work out of the box — no API keys needed.