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

# Get Started

## What is Munsit Speech to Text?

Munsit Speech to Text is an advanced AI-powered service that converts spoken Arabic into accurate, structured text. It is built for high-accuracy Arabic recognition across dialects and real-world speech conditions, helping teams turn audio into searchable and actionable information.

### Key Features

* **High-Accuracy Arabic Transcription**: Transcribes spoken Arabic with strong performance across dialects and accents
* **Speaker Diarization**: Separates and labels speakers for clearer multi-speaker conversations
* **Meeting Intelligence Ready**: Supports minutes of meetings and downstream analysis workflows
* **Flexible Audio Support**: Works with common audio formats from different recording platforms

Munsit Speech to Text helps you process interviews, meetings, calls, and media content at scale while keeping Arabic language quality and context at the center.

## Core Workflows

* **Transcriber**: Convert pre-recorded audio files into text with word-level timing
* **Diarization**: Identify who spoke and return speaker-labeled transcript segments
* **Minutes of Meetings**: Generate structured transcripts optimized for meeting use cases

## Available Models

| Model        | ID             | Description                                                               |
| ------------ | -------------- | ------------------------------------------------------------------------- |
| Munsit       | `munsit`       | Default model, optimized for Arabic speech recognition                    |
| Munsit En-Ar | `munsit-en-ar` | Model for mixed Arabic-English spoken content with code-switching support |

All API endpoints accept an optional `model` parameter. If omitted, `munsit` is used by default.

## Supported Language

Munsit is optimized for Arabic speech recognition and transcription, with strong coverage for dialect and accent variation. The `munsit-en-ar` model is designed for mixed Arabic-English spoken content, supporting code-switching where speakers naturally alternate between the two languages within the same conversation or utterance.

## Supported Audio Formats

`.aac`, `.wma`, `.amr`, `.flac`, `.m4a`, `.ogg`, `.mp2`, `.opus`, `.m4r`, `.webm`, `.mp3`, `.wav`

## Duration Limits

* **Audio Transcription**: Supports files shorter than 60 minutes
* **Minutes of Meetings**: Supports files shorter than 30 minutes
* **Streaming**: No duration limit while the connection stays active

For longer recordings in Audio Transcription or Minutes of Meetings, split the audio into shorter segments for best performance.