Whisper-small-el Open-source Speech Recognition Model - Free Deployment, Precise Recognition of Greek Speech

Whisper Small El

Developed by mozilla-ai

This is an automatic speech recognition (ASR) model fine-tuned from the openai/whisper-small model for Greek speech recognition tasks, trained on 3,620 Greek samples from the Mozilla Common Voice 17.0 dataset.

Speech Recognition

Transformers

OtherOpen Source License:Apache-2.0 #Greek speech-to-text #Low-resource optimization #Multi-scenario ASR

Downloads 94

Release Time : 2/17/2025

Model Overview

This model is specifically designed for Greek speech-to-text tasks, achieving a word error rate (WER) of 45.632 on the Common Voice Greek test set.

Model Features

Greek optimization

Specially fine-tuned for Greek, with improved recognition accuracy compared to the original model

Open-source dataset training

Trained using the Mozilla Common Voice open-source dataset with reliable data sources

Performance improvement

Compared to the baseline model, the word error rate decreased from 46.392 to 45.632, with improved loss values

Model Capabilities

Greek speech recognition

Speech-to-text

Audio transcription

Use Cases

Speech transcription

Greek meeting minutes

Automatically transcribe Greek meeting recordings into text records

Word error rate approximately 45.6%

Greek media subtitle generation

Automatically generate subtitles for Greek video content

Approximately 35.7% WER on FLEURS test set

Voice assistant

Greek voice command recognition

Used for voice command understanding in Greek voice assistant systems

Property	Details
Base Model	openai/whisper-small
Datasets	mozilla-foundation/common_voice_17_0
Language	el
Library Name	transformers
License	apache-2.0

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

Whisper Small El

Model Overview

Model Features

Model Capabilities

Use Cases

🚀 Finetuned openai/whisper-small on Greek

🚀 Quick Start

📚 Documentation

Model Information

Evaluation Results

Evaluation results on 1701 audio samples of Greek:

Baseline model (before finetuning) on Greek

Finetuned model (after finetuning) on Greek

Finetuned model (after finetuning) on the Greek FLEURS test set (total of 649 samples)

📄 License