# Advanced (/agents/advanced)





## Automatic Turn Detection

Controls **how the Agent knows when a user has finished speaking** and when it should respond. <br />

### Semantic

The Agent listens for **meaning and intent**, not just silence.  <br />
It continuously tries to understand if the user has finished speaking, even if the user pauses. <br />
**Example:**

> “I was wondering if you could help me book a demo… maybe tomorrow?” <br />
> 👉 The agent waits for the full thought before responding.

Note that you can even control **how eager** is the Agent to respond.

<img alt="AgentAdvanced" src={__img0} placeholder="blur" />

### Standard (Timeout-Based)

The Agent responds after a **pause or silence** based on a fixed timeout (e.g., 1–2 seconds of no speech)

### When to Use Each

* **Semantic** → More natural, human-like conversations (recommended)
* **Standard** → Simpler use cases or when timing must be strictly controlled

## Noise Reduction

Improves how well the Agent hears the user by filtering background noise.

### Near-Field

Designed for users speaking **close to the microphone** as phone calls, headset, laptop mic.<br />
👉 Best for most use cases

### Far-Field

Designed for users speaking **from a distance** as speakerphones, meeting rooms, kiosks.<br />
👉 Helps capture speech in noisier environments

### When to Use Each

* **Near-field** → One-on-one calls (default)
* **Far-field** → Shared spaces or noisy environments

<img alt="AgentAdvanced2" src={__img1} placeholder="blur" />

## Transcription

There settings control how the Agent converts speech into text.

### Automatic Language Detection

The Agent automatically detects the spoken language out of **50+ supported**. <br />
This is best when the users are expected to speak different languages, however it comes with a drawback in accuracy, especially with shorted sentences.

### Preselected Language

You define a specific language (e.g., English) which improves accuracy even for shorter sentences.

Note that you can even choose from **high precision** or **high accuracy**

### Additional Instructions

You can provide extra guidance to improve transcription accuracy.

### Examples

> * “Expect email addresses and phone numbers”
> * “Support technical terms and abbreviations”

### Why This Matters

Better transcription leads to more accurate understanding, better tool usage and more natural responses.