# Fireworks

> Fast AI inference with Fireworks AI models, including audio transcription with speaker diarization using Whisper models.

Source: https://cotera.co/docs/reference/tools/individual-tools/fireworks-ai

---

**Authentication Type:** API Key\
**Description:** Fast AI inference with Fireworks AI models, including audio transcription with speaker diarization using Whisper models.

***

## Authentication

To authenticate, you'll need a Fireworks AI API key. Learn how to create one in the [Fireworks AI documentation](https://fireworks.ai/api-keys).

***

## Whisper

Audio transcription with speaker diarization using Whisper models.

### Transcribe Audio

Transcribe audio from a URL with speaker diarization. Returns timestamped segments with speaker labels.

**Operation Type:** Query (Read)

**Parameters:**

* **audioUrl** `string` (required): URL to the audio file to transcribe (max 1GB)
* **language** `string` (nullable): Target language for transcription (ISO-639-1 format, e.g., "en", "es"). If not provided, language will be auto-detected.
* **prompt** `string` (nullable): Optional prompt to guide transcription style. E.g., "Um, here's, uh, what was recorded." to include filler words.
* **model** `string` (nullable): Model to use. Options: "whisper-v3-turbo" (default) for fast processing, "whisper-v3" for best quality.
* **minSpeakers** `number` (nullable): Minimum number of speakers to detect for diarization
* **maxSpeakers** `number` (nullable): Maximum number of speakers to detect for diarization

**Returns:**

* **segments** `array of objects`: Array of transcribed segments with speaker diarization
  * **id** `number`: Segment ID
  * **start** `number`: Start timestamp in seconds
  * **end** `number`: End timestamp in seconds
  * **text** `string`: Transcribed text for this segment
  * **speakerId** `string` (nullable): Speaker identifier (e.g., "SPEAKER\_00", "SPEAKER\_01")
* **text** `string` (nullable): Full transcribed text
* **duration** `number` (nullable): Duration of the audio in seconds
* **language** `string` (nullable): Detected language of the audio

**Example Usage:**

```json
{
  "audioUrl": "https://example.com/audio/meeting-recording.mp3",
  "language": "en",
  "model": "whisper-v3",
  "minSpeakers": 2,
  "maxSpeakers": 5
}
```

***

## Common Use Cases

**Meeting Transcription:**

* Transcribe recorded meetings with speaker identification
* Generate meeting notes with timestamps and speaker labels
* Create searchable archives of recorded conversations

**Content Creation:**

* Transcribe podcasts and video content for accessibility
* Generate subtitles with accurate speaker attribution
* Convert audio interviews into written content

**Analysis & Research:**

* Transcribe customer calls for quality analysis
* Process research interviews with speaker segmentation
* Analyze multi-speaker audio content for insights