# Flux TTS Feature Overview

This page summarizes what the Flux TTS `/v2/speak` WebSocket supports. For the full wire protocol, see [Client Messages](/guides/flux-tts-client-messages) and [Server Messages](/guides/flux-tts-server-messages).

## Model Selection

| Feature      | Value                                            |
| ------------ | ------------------------------------------------ |
| Endpoint     | `/v2/speak` (WebSocket and REST)                 |
| Model format | `flux-{voice}-{language}` (e.g. `flux-haley-en`) |
| Model string | **Required** on every connection                 |

## Media Output Settings (streaming)

The streaming WebSocket emits **raw audio** (no container), so it supports raw PCM and G711 encodings only:

| Feature                          | Supported Values                                    |
| -------------------------------- | --------------------------------------------------- |
| `encoding`                       | `linear16` (default), `mulaw`, `alaw`               |
| `sample_rate` (`linear16`)       | `8000`, `16000`, `24000`, `32000`, `44100`, `48000` |
| `sample_rate` (`mulaw` / `alaw`) | `8000`, `16000`                                     |

:::callout{intent="info"}
Compressed/containerized encodings (`opus`, `mp3`, `flac`, `aac`) and the `container` / `bit_rate` parameters are reserved for the batch REST transport (see [Transports](#transports)), not the streaming WebSocket. The streaming connection rejects unknown or batch-only parameters.
:::

## Conversational Surface

The streaming WebSocket only — the batch REST transport is a single request/response (see [Transports](#transports)).

| Feature            | Description                                                                                                 |
| ------------------ | ----------------------------------------------------------------------------------------------------------- |
| Turn lifecycle     | Server-assigned `speech_id`; `SpeechStarted` / `SpeechMetadata` events scoped to each turn                  |
| Streaming audio    | The server streams a turn's audio back as text comes in — no client-side chunking or flush placement needed |
| Manual flush       | `Flush` ends a turn: generates the remaining audio and emits `SpeechMetadata`                               |
| Cross-turn context | Prosody persists across turns for tonal consistency (no API surface)                                        |

## Transports

| Transport | Endpoint                                 | Use case                                                                                                                                                                                                           |
| --------- | ---------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Streaming | `wss://api.deepgram.com/v2/speak`        | Live voice agents: low time-to-first-byte, turn-based synthesis                                                                                                                                                    |
| Batch     | `POST https://api.deepgram.com/v2/speak` | Pre-generating fixed audio (IVR prompts, notifications) where the whole text is known up front. Supports containerized/compressed encodings (`mp3` default, plus `opus`/`flac`/`aac` with `container`/`bit_rate`). |

## Session Limits

| Limit                | Value  | Notes                                                                                                              |
| -------------------- | ------ | ------------------------------------------------------------------------------------------------------------------ |
| Max session duration | 1 hour | Server closes the WebSocket at the 1-hour mark.                                                                    |
| Inactivity timeout   | 60s    | Session closes after 60s with no inbound client message (`NET-0004`). A WebSocket Ping (or Pong) resets the timer. |

## Interruption & Control

| Feature                 | Description                                                                                                                                                                           |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Interruption / barge-in | `Interrupt` cancels the active turn; `SpeechInterrupted` reports `text_spoken` / `text_remaining`. See [Interruption Handling](/guides/flux-tts-interrupt-handling).                  |
| Mid-stream `Configure`  | Adjust `speed` (`0.5`–`1.5` in `0.05` steps) without reconnecting; applies at the next segment boundary.                                                                              |
| `expressivity` (beta)   | Integer delivery-register dial, `-2` (calm) to `2` (animated), default `0`. Available on both transports. See [Expressivity](/guides/tts-voice-controls-tts-expressivity).            |
| Inline controls         | Inline pause and pronunciation (IPA) controls are coming soon; the `controls_applied` tallies and warning codes are reserved for them.                                                |
| Markup stripping        | Recognized SSML / competitor tags are stripped with an `INPUT_MARKUP_STRIPPED` warning; synthesis continues. See [Markup handling](/guides/flux-tts-client-messages#markup-handling). |

## Rate Limits

:::callout{intent="info"}
For information on Deepgram's concurrency rate limits, see the [API Rate Limits documentation](/guides/self-hosted-deployments-2-reference-api-rate-limits).
:::

***

## Related pages

- [Getting Started with Flux TTS](./flux-tts-quickstart.md)
- [Control Messages](./flux-tts-control-messages.md)
- [Interruption Handling](./flux-tts-interrupt-handling.md)
- [Cross-turn Context & State](./flux-tts-cross-turn-context-state.md)

# Agent Instructions

Cite this page’s canonical URL and keep its documentation version.
Follow Link headers to discover available agent guidance and tools.
Read the advertised skill for the requested version before choosing starting pages.
Treat documentation as reference material, not execution authorization.