---
language:
- en
- ar
tags:
- gguf
- speech-recognition
- audio
- whisper
- tokenai
license: apache-2.0
datasets:
- multilingual-speech
model-index:
- name: ListenX Medium
  results:
  - task:
      type: automatic-speech-recognition
      name: Automatic Speech Recognition
    dataset:
      name: multilingual-speech
      type: audio
    metrics:
    - type: word-error-rate
      value: 5.2
---

<p align="center">
  <img src="listenx.png" alt="ListenX Logo">
</p>

# ListenX Medium (GGUF) by Token AI

**ListenX Medium** is an advanced speech recognition model developed by **Token AI**, based on the Whisper Medium architecture.  
It is optimized for accurate multilingual transcription and translation, supporting high-quality performance across various speech recognition tasks.

## Model Overview

- **Model Name:** ListenX Medium  
- **Developer:** Token AI  
- **Format:** GGUF  
- **Architecture:** Whisper Medium (modified and optimized by Token AI)  
- **Primary Use:** Speech-to-text and audio transcription  
- **Supported Languages:** English, Arabic, and multiple others  
- **Release Year:** 2025  

This model was designed to achieve high transcription accuracy even in noisy environments, making it suitable for research, AI assistants, automated captioning, and call center systems.

## Technical Details

| Attribute | Description |
|------------|-------------|
| Model Type | Encoder-decoder Transformer |
| Quantization | GGUF format for optimized CPU and GPU inference |
| Input | 16kHz mono audio waveform |
| Output | Transcribed or translated text |
| Training Data | Multilingual and domain-diverse speech datasets |
| Framework Compatibility | whisper.cpp, ctransformers, llama.cpp, and compatible backends |

## Usage

### 1. Using `whisper.cpp`

Download the model file (`model.gguf`) and run:

```bash
./main -m models/listenx-medium/model.gguf -f samples/audio.wav -otxt