# Speech-to-speech architecture

Supported in ADKPython v0.1.0Java v0.2.0

In a speech-to-speech architecture, a single multimodal model processes incoming audio, executes reasoning and tool calls, and generates spoken audio directly over a persistent [Gemini Live API](https://ai.google.dev/gemini-api/docs/live-api) connection, without converting the speech to text between the input and output..

For a comparison with chained STT-LLM-TTS pipelines, see [Choose a voice architecture](https://adk.dev/live/architectures/index.md).

## Configure a speech-to-speech agent

Pass a supported [native audio Live model](https://adk.dev/live/models/#live-models) to your agent:

```python
import os
from google.adk.agents import Agent

root_agent = Agent(
    name="voice_assistant",
    model=os.getenv("DEMO_AGENT_MODEL", "gemini-live-2.5-flash-native-audio"),
    instruction="You are a helpful voice assistant.",
    tools=[...],
)
```

```java
import com.google.adk.agents.LlmAgent;

LlmAgent rootAgent =
    LlmAgent.builder()
        .name("voice_assistant")
        .model("gemini-live-2.5-flash-native-audio")
        .instruction("You are a helpful voice assistant.")
        .build();
```

Run the agent with `adk web` or pass it to [`Runner.run_live()`](https://adk.dev/live/sessions/#the-run_live-loop).

## Capabilities specific to speech-to-speech

The following capabilities apply specifically to native audio models:

- **Live video and image input**: Stream live camera or screen frames alongside audio so the agent can see what the user is showing. See [Audio and video](https://adk.dev/live/audio-video/#images-and-video).
- **Affective dialog and proactive audio**: Supported native audio models can adapt their spoken tone to the user's emotional expression and decide when to speak or stay silent. See [Proactivity and affective dialog](https://adk.dev/live/configuration/#proactivity-and-affective-dialog).
- **Stateful session resumption and context compression**: Because the Live API holds conversation state in a stateful WebSocket session, ADK supports automatic [session resumption](https://adk.dev/live/sessions/#session-resumption) across connection timeouts and [context window compression](https://adk.dev/live/sessions/#context-window-compression) for long-running sessions.
- **Session-level voice and VAD controls**: Configure voice selection and automatic voice activity detection (VAD) at the session level through [`RunConfig`](https://adk.dev/live/configuration/index.md) or on the model instance.

## Considerations and limitations

- **Fixed model selection**: Speech-to-speech requires a model that supports the Gemini Live API ([Supported models](https://adk.dev/live/models/#live-models)). To use standard Gemini text models or third-party LLMs in a live voice session, use the [Cascaded architecture](https://adk.dev/live/architectures/cascade/index.md).
- **Audio-only output**: Native audio models output audio only and do not run under `Runner.run_async()`. Read [audio transcriptions](https://adk.dev/live/configuration/#audio-transcription) when you need the text of the conversation.
- **Guardrails evaluate transcriptions alongside audio**: Because the model generates audio directly without an intermediate text stage, [conversation validation callbacks](https://adk.dev/live/guardrails/#conversation-validation) screen input and output transcriptions rather than intercepting text before synthesis. If you must filter or redact text before audio is generated, use [Cascade](https://adk.dev/live/architectures/cascade/#custom-speech-transforms).