Hamsa-v0.1-beta Open-Source Arabic Speech Recognition Model - Meeting the Language Needs of the Middle East and North Africa

Hamsa V0.1 Beta

Developed by nadsoft

Hamsa is an Arabic speech recognition model built on the Whisper architecture, focusing on the linguistic needs of the Middle East and North Africa region.

Speech Recognition

Transformers

ArabicOpen Source License:Apache-2.0 #Arabic speech recognition #Whisper architecture optimization #MENA region adaptation

Downloads 46

Release Time : 10/20/2023

Model Overview

Hamsa (همسة) is a pre-trained Arabic Automatic Speech Recognition (ASR) model dedicated to addressing the unique linguistic nuances of Arabic, particularly suited for use in the Middle East and North Africa region.

Model Features

Arabic optimization

Specially optimized for Arabic to address the unique linguistic nuances of the Middle East and North Africa region.

Multi-source data training

Trained using multiple Arabic speech datasets, including nadsoft/Jordan-Audio, google/fleurs, and common_voice_11_0.

Technologically advanced

Utilizes the Whisper model architecture with advanced techniques such as Adam optimizer and mixed-precision training.

Model Capabilities

Arabic speech recognition

Standard Arabic transcription

Speech-to-text

Use Cases

Speech transcription

Standard Arabic transcription

Convert standard Arabic speech into text

Word Error Rate (WER) of 18.22

Education

Arabic learning assistance

Helps learners practice Arabic pronunciation and transcription

🚀 Hamsa-v0.1-beta

Hamsa is a pre - trained Arabic automatic speech recognition model based on the Whisper model, aiming to improve AI performance for the Arabic language.

🚀 Quick Start

There is no specific quick - start content in the original document, so this section is skipped.

✨ Features

Hamsa (همسة) represents a sophisticated advancement in the realm of Arabic speech recognition. It's a pre - trained automatic speech recognition (ASR) model that is built upon the foundation of the Whisper model. Hamsa is not just a technological achievement; it's a testament to NADSOFT's commitment to elevating the standards of AI results for the Arabic language. This contribution is especially significant for the Middle East and North Africa (MENA) region and the broader Arab World, as it seeks to address the unique linguistic nuances and cater to the specific needs of these communities.

📚 Documentation

Intended uses & limitations

Hamsa is a model that is still under development, and it is important to be aware of its limitations. For example, the model may not be able to accurately transcribe text from speakers with very strong accents, such as Moroccan Arabic. Additionally, the model may have difficulty transcribing text from noisy recordings.

It is important to note that Hamsa is not a perfect model, and it should not be used to generate text that is intended to be used in legal, medical, or other sensitive contexts.

Training and evaluation data

The training and evaluation data includes nadsoft/Jordan - Audio, google/fleurs, and mozilla - foundation/common_voice_11_0. The Word Error Rate (WER) is 18.22.

Training procedure

Training hyperparameters

learning_rate: 1e - 05
train_batch_size: 32
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 10000 then 4000 for NADSOFT data
mixed_precision_training: Native AMP

📄 License

This model is licensed under the Apache - 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご