# Speko ## Docs - [Audio helpers](https://docs.speko.dev/adapter-livekit/audio.md): WAV encode / decode and MIME parsing utilities. - [createSpekoComponents](https://docs.speko.dev/adapter-livekit/create-speko-components.md): Build a { stt, llm, tts } bundle ready for voice.AgentSession. - [Intent](https://docs.speko.dev/adapter-livekit/intent.md): Routing hint type and construction-time validator. - [@spekoai/adapter-livekit](https://docs.speko.dev/adapter-livekit/overview.md): LiveKit Agents adapter — route STT, LLM, and TTS through Speko. - [SpekoLLM](https://docs.speko.dev/adapter-livekit/speko-llm.md): LiveKit Agents LLM adapter backed by POST /v1/complete. - [SpekoSTT](https://docs.speko.dev/adapter-livekit/speko-stt.md): LiveKit Agents STT adapter backed by POST /v1/transcribe. - [SpekoTTS](https://docs.speko.dev/adapter-livekit/speko-tts.md): LiveKit Agents TTS adapter backed by POST /v1/synthesize. - [spekoai-mcp](https://docs.speko.dev/ai-tools/mcp.md): Model Context Protocol server exposing Speko docs, scaffolds, and balance tools to Claude, Cursor, and other MCP clients. - [Complete](https://docs.speko.dev/api-reference/complete.md): Single-turn LLM call routed to the best provider for your intent, with automatic failover. Returns assistant text + token usage. - [Introduction](https://docs.speko.dev/api-reference/introduction.md): Public REST API for the Speko voice gateway. - [Sessions](https://docs.speko.dev/api-reference/sessions.md): Mints a LiveKit access token, persists the pipeline config, and dispatches an agent worker into the room. Use the returned `conversationToken` and `livekitUrl` with `@spekoai/client` to join the room from a browser. - [Synthesize](https://docs.speko.dev/api-reference/synthesize.md): Routes the request to the best TTS provider for your intent, with automatic failover. Returns binary audio. The `Content-Type` header (mirrored in `X-Speko-Audio-Format`) tells you the format. - [Transcribe](https://docs.speko.dev/api-reference/transcribe.md): Routes the request to the best STT provider for your `(language, optimizeFor)` intent, with automatic failover to runner-up providers. Body is binary audio. Routing intent goes in the `x-speko-intent` header (JSON). - [Callbacks & events](https://docs.speko.dev/client/callbacks.md): Every hook VoiceConversation exposes, and when they fire. - [Data channel protocol](https://docs.speko.dev/client/data-channel.md): Wire format for packets exchanged between browser and agent over LiveKit data channels. - [Errors](https://docs.speko.dev/client/errors.md): SpekoClientError and its error codes. - [@spekoai/client](https://docs.speko.dev/client/overview.md): Browser SDK for real-time voice conversations over LiveKit WebRTC. - [VoiceConversation](https://docs.speko.dev/client/voice-conversation.md): Primary API — construct, control, and tear down a voice session. - [Bring your own keys](https://docs.speko.dev/concepts/byok.md): Use your own provider credentials. Speko routes; providers bill you directly. - [Failover](https://docs.speko.dev/concepts/failover.md): How Speko transparently retries against runner-up providers when the primary fails. - [How routing works](https://docs.speko.dev/concepts/routing.md): Intent + benchmark scores → ranked candidates → live failover. The model behind every Speko call. - [Benchmarks and scoring](https://docs.speko.dev/concepts/scoring.md): How provider scores are computed and refreshed. - [Sessions vs one-shot](https://docs.speko.dev/concepts/sessions.md): When to mint a session and when to call a one-shot endpoint directly. - [API keys](https://docs.speko.dev/dashboard/api-keys.md): Mint, list, and revoke organization API keys. - [Bring your own keys](https://docs.speko.dev/dashboard/byok.md): Connect your own provider API keys so providers bill you directly while Speko still routes. - [Team and organization](https://docs.speko.dev/dashboard/team.md): Invite teammates, manage members, configure your organization. - [Tools](https://docs.speko.dev/dashboard/tools.md): Register webhook tools your voice agents can invoke mid-call. - [Usage](https://docs.speko.dev/dashboard/usage.md): Per-provider, per-metric usage broken down by date range. - [One-shot APIs](https://docs.speko.dev/guides/one-shot.md): POST /v1/transcribe, /v1/synthesize, /v1/complete — single-turn calls without sessions. - [Real-time browser conversation](https://docs.speko.dev/guides/realtime-conversation.md): Wire @spekoai/client into a web app — mint a session, join the room, stream voice both ways. - [Tool calling](https://docs.speko.dev/guides/tool-calling.md): Give a Speko voice agent the ability to invoke webhook tools mid-call. Register once in the dashboard, fire from any voice session. - [Build a voice agent](https://docs.speko.dev/guides/voice-agent.md): Real-time STT → LLM → TTS pipeline using @spekoai/adapter-livekit on a LiveKit Agents worker. - [Speko](https://docs.speko.dev/index.md): One API for every voice provider. Speko benchmarks STT, LLM, and TTS in real time and routes each call to the best provider for your language and latency/cost target. - [Quickstart](https://docs.speko.dev/quickstart.md): From sign-up to your first transcribe call in under five minutes. - [complete](https://docs.speko.dev/sdk-python/complete.md): POST /v1/complete — LLM completion with automatic provider routing. - [credits](https://docs.speko.dev/sdk-python/credits.md): Prepaid credit balance and append-only ledger. - [Errors](https://docs.speko.dev/sdk-python/errors.md): Typed exceptions raised by the Python SDK. - [spekoai (Python)](https://docs.speko.dev/sdk-python/overview.md): Official Python SDK for the Speko voice gateway — sync + async. - [connect_realtime](https://docs.speko.dev/sdk-python/realtime.md): Speech-to-speech WebSocket sessions, async only. - [synthesize](https://docs.speko.dev/sdk-python/synthesize.md): POST /v1/synthesize — text-to-speech with automatic provider routing. - [transcribe](https://docs.speko.dev/sdk-python/transcribe.md): POST /v1/transcribe — speech-to-text with automatic provider routing. - [usage](https://docs.speko.dev/sdk-python/usage.md): GET /v1/usage — billing and usage reporting. - [complete](https://docs.speko.dev/sdk/complete.md): POST /v1/complete — LLM completion with automatic provider routing. - [credits](https://docs.speko.dev/sdk/credits.md): Prepaid balance and append-only ledger. - [Errors](https://docs.speko.dev/sdk/errors.md): Exception classes thrown by @spekoai/sdk. - [@spekoai/sdk](https://docs.speko.dev/sdk/overview.md): Official TypeScript SDK — one API, every voice provider. - [realtime](https://docs.speko.dev/sdk/realtime.md): Speech-to-speech WebSocket sessions — OpenAI Realtime, Gemini Live, xAI Grok Voice. - [Speko client](https://docs.speko.dev/sdk/speko-client.md): Construct the SDK client and configure transport options. - [synthesize](https://docs.speko.dev/sdk/synthesize.md): POST /v1/synthesize — text-to-speech with automatic provider routing. - [transcribe](https://docs.speko.dev/sdk/transcribe.md): POST /v1/transcribe — speech-to-text with automatic provider routing. - [Types](https://docs.speko.dev/sdk/types.md): Shared request / response types exported from @spekoai/sdk. - [usage](https://docs.speko.dev/sdk/usage.md): GET /v1/usage — billing and usage reporting. ## OpenAPI Specs - [openapi](https://docs.speko.dev/api-reference/openapi.json) ## Optional - [Website](https://speko.ai) - [Dashboard](https://speko.dev)