# .wave > APIs for real-time speech: full-duplex voice conversations with the Live API, and streaming transcription with the Realtime API and the Deepgram-compatible API. Each documentation page is also available as Markdown, at the addresses below. ## Get started - [.wave documentation](https://dotwave.ai/docs/index.md): Build with the .wave APIs for real-time speech: the Live API for full-duplex voice conversations, the Realtime API for streaming transcription in the OpenAI format, and the Deepgram-compatible API for Deepgram clients. - [Quickstart](https://dotwave.ai/docs/quickstart/index.md): Create a .wave API key, check it, and make a first call to the Live API, the Realtime API or the Deepgram-compatible API. - [Authentication](https://dotwave.ai/docs/authentication/index.md): Authenticate to .wave with an API key from your server, or with a short-lived client secret from a browser. - [Errors](https://dotwave.ai/docs/errors/index.md): The .wave error envelope, status codes, request IDs and retries, and where each API documents its socket errors and close codes. - [Pricing and limits](https://dotwave.ai/docs/pricing-and-limits/index.md): Prices per model, free credits, session limits and project limits on .wave. ## LLM API - [Gemma 4 31B](https://dotwave.ai/docs/models/google/gemma-4-31b-it/index.md): Stream text and function calls with Gemma 4 31B through the Wave API. - [LLM chat completions](https://dotwave.ai/docs/guides/llm/index.md): Use Python, TypeScript and LiveKit Agents to stream Gemma responses with conversation affinity. ## Live API - [Live API](https://dotwave.ai/docs/live/index.md): The .wave Live API: full-duplex speech-to-speech sessions over one WebSocket, with the model’s speech and transcripts of both sides. - [Live API client secrets](https://dotwave.ai/docs/live/client-secrets/index.md): Create a client secret with POST /v1/live/client_secrets so a browser can open one Live API session without your API key. - [Live API WebSocket](https://dotwave.ai/docs/live/websocket/index.md): Connect to the Live API WebSocket, start a session with session.start, stream 24 kHz PCM16 audio both ways, and handle close codes. - [Live API client events](https://dotwave.ai/docs/live/client-events/index.md): The events a client sends on the Live API WebSocket: session.start, session.input_audio.append, mute, update and close, and the events it refuses. - [Live API server events](https://dotwave.ai/docs/live/server-events/index.md): The events the Live API WebSocket sends: session state, model speech, user and model transcripts, turn events, usage, the close and errors. ## Realtime API - [Realtime API](https://dotwave.ai/docs/realtime/index.md): The .wave Realtime API: streaming transcription in the OpenAI Realtime transcription format. The OpenAI SDK connects by changing its base URL. - [Realtime API client secrets](https://dotwave.ai/docs/realtime/client-secrets/index.md): Create a client secret with POST /v1/realtime/client_secrets so a browser can open one transcription session without your API key. - [Realtime API WebSocket](https://dotwave.ai/docs/realtime/websocket/index.md): Connect to the Realtime API WebSocket, configure a transcription session, stream audio at the speed of speech, and handle segments and close codes. - [Realtime API client events](https://dotwave.ai/docs/realtime/client-events/index.md): The events a client sends on the Realtime API WebSocket: session.update and the input_audio_buffer events, with the transcription session fields. - [Realtime API server events](https://dotwave.ai/docs/realtime/server-events/index.md): The events the Realtime API WebSocket sends: session.created, session.updated, speech events, transcription deltas and completed segments, errors and session.end. - [Transcription languages](https://dotwave.ai/docs/realtime/languages/index.md): Name the language of the audio for the .wave Realtime API and Deepgram-compatible API, or let the model detect it. ## Deepgram-compatible API - [Deepgram-compatible API](https://dotwave.ai/docs/deepgram/index.md): The .wave Deepgram-compatible API: streaming transcription on wss://api.dotwave.ai/v1/listen for Deepgram clients, including the LiveKit Agents and Pipecat plugins. - [Deepgram-compatible API parameters](https://dotwave.ai/docs/deepgram/parameters/index.md): The query parameters of the .wave Deepgram-compatible socket: encoding, sample_rate, language, utterance_end_ms, interim_results and the parameters it accepts or refuses. - [Deepgram-compatible API messages](https://dotwave.ai/docs/deepgram/messages/index.md): The messages of the .wave Deepgram-compatible socket: Metadata, Results with whole words, SpeechStarted and UtteranceEnd, KeepAlive, Finalize and CloseStream, errors and close codes. - [LiveKit Agents and Pipecat](https://dotwave.ai/docs/deepgram/plugins/index.md): Use .wave speech-to-text in Pipecat with the pipecat-dotwave service, and in LiveKit Agents through its Deepgram plugin with the base URL changed. - [Switch from Deepgram](https://dotwave.ai/docs/deepgram/switch/index.md): Move a Deepgram streaming transcription client to .wave: change the base URL and the key, keep the code. Examples for the Deepgram SDKs, LiveKit Agents, Pipecat, Vapi and a plain WebSocket, with what carries over and what changes. ## Reference - [Model catalogue](https://dotwave.ai/docs/models/index.md): Models available on .wave, and the API that serves each: full-duplex voice on the Live API, streaming transcription on the Realtime API and the Deepgram-compatible API. - [Nemotron 3.5 ASR Streaming 0.6B](https://dotwave.ai/docs/models/nvidia/nemotron-asr-streaming-0-6b/index.md): Run NVIDIA Nemotron 3.5 ASR Streaming 0.6B on the .wave Realtime API and Deepgram-compatible API, in public beta: streaming transcription in 32 language locales with the OpenAI SDK or a Deepgram plugin. - [Nemotron VoiceChat 11B](https://dotwave.ai/docs/models/nvidia/nemotron-voicechat-11b/index.md): Run NVIDIA NemotronLabs VoiceChat 11B on the .wave Live API, available on request: full-duplex speech to speech. - [API reference](https://dotwave.ai/docs/api/index.md): Reference for the .wave APIs, grouped by API: the Live API, the Realtime API, the Deepgram-compatible API, and the models, sessions and usage routes they share. ## Guides - [Python](https://dotwave.ai/docs/guides/python/index.md): Stream microphone audio over a plain WebSocket on the .wave Live API and play the model’s speech, in Python. - [TypeScript](https://dotwave.ai/docs/guides/typescript/index.md): Stream microphone audio over a plain WebSocket on the .wave Live API and receive the model’s speech and transcripts, in TypeScript. - [OpenAI SDK](https://dotwave.ai/docs/guides/openai-sdk/index.md): Transcribe with the OpenAI SDK for Python or TypeScript on the .wave Realtime API, changing only the base URL and the key.