Daniel_asr Open-source Automatic Speech Recognition Model - Free Deployment, with Word Error Rate as Low as 0.3423 on Evaluation Set

Daniel Asr

Developed by danielbubiola

An automatic speech recognition (ASR) model fine-tuned from facebook/wav2vec2-base, achieving a word error rate of 0.3423 on the evaluation set

Speech Recognition

Transformers

Open Source License:Apache-2.0 #Speech Recognition #Low Word Error Rate #Wav2Vec2 Fine-tuning

Downloads 25

Release Time : 3/2/2022

Model Overview

This is a model for automatic speech recognition (ASR), fine-tuned based on Facebook's wav2vec2-base architecture, capable of converting speech to text

Model Features

Efficient Speech Recognition

Based on the wav2vec2 architecture, providing efficient speech-to-text capabilities

Low Word Error Rate

Achieved a word error rate (WER) of 0.3423 on the evaluation set

Fine-tuning Optimization

Performed 30 epochs of fine-tuning on the base model, significantly improving performance

Model Capabilities

Speech-to-text

Automatic Speech Recognition

Use Cases

Speech Transcription

Meeting Minutes Transcription

Automatically convert meeting recordings into text transcripts

Word error rate 0.3423

Voice Note Conversion

Convert voice memos into editable text

Training Loss	Epoch	Step	Validation Loss	Wer
3.4909	4.0	500	1.3485	0.8887
0.5887	8.0	1000	0.4957	0.4641
0.2207	12.0	1500	0.4621	0.3971
0.125	16.0	2000	0.4339	0.3756
0.0829	20.0	2500	0.4618	0.3613
0.0601	24.0	3000	0.4564	0.3535
0.0456	28.0	3500	0.4565	0.3423

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

Daniel Asr

Model Overview

Model Features

Model Capabilities

Use Cases

🚀 daniel_asr

🚀 Quick Start

📚 Documentation

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

📄 License