> ## Documentation Index
> Fetch the complete documentation index at: https://openrouter.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# STTWord

A timestamped word, returned when the provider includes word-level timestamps

## Example Usage

```typescript theme={null}
import { STTWord } from "@openrouter/sdk/models";

let value: STTWord = {
  end: 0.4,
  start: 0,
  word: "Hello",
};
```

## Fields

| Field | Type | Required | Description | Example |
| - | - | - | - | - |
| `channel` | *number* | :heavy\_minus\_sign: | Zero-based audio channel index for the word, present when the provider transcribes channels separately | 0 |
| `confidence` | *number* | :heavy\_minus\_sign: | Provider confidence for the word from 0 to 1, present when the provider returns per-word confidence | 0.98 |
| `end` | *number* | :heavy\_check\_mark: | Word end time in seconds | 0.4 |
| `speaker` | *number* | :heavy\_minus\_sign: | Speaker index for the word, present when the provider returns diarization data | 0 |
| `speakerLabel` | *string* | :heavy\_minus\_sign: | Provider speaker label for the word, present when the provider labels speakers with a string | speaker\_0 |
| `start` | *number* | :heavy\_check\_mark: | Word start time in seconds | 0 |
| `type` | [models.STTWordType](../models/sttwordtype.mdx) | :heavy\_minus\_sign: | Kind of entry; omitted or "word" for spoken words, "audio\_event" for non-speech sounds the provider tags with timestamps | word |
| `word` | *string* | :heavy\_check\_mark: | The transcribed word, or the event tag such as "(laughter)" when type is audio\_event | Hello |