Librispeech-100h-supervised-meta Open-source Speech Recognition Model

Librispeech 100h Supervised Meta

Developed by Kuray107

Speech recognition model fine-tuned from Kuray107/librispeech-5h-supervised, trained on 100 hours of LibriSpeech dataset

Speech Recognition

Transformers

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

Downloads 25

Release Time : 4/8/2022

Model Overview

This model is a supervised learning-based speech recognition model, specifically optimized for English speech recognition tasks, capable of converting speech to text.

Model Features

High Accuracy

Achieved an excellent word error rate (WER) of 0.033 on the evaluation set

Supervised Learning

Trained using supervised learning methods with annotated data for optimization

LibriSpeech Dataset

Fine-tuned on 100 hours of LibriSpeech English speech dataset

Model Capabilities

English Speech Recognition

Speech-to-Text

Use Cases

Speech Transcription

Meeting Minutes

Automatically transcribe meeting recordings into text records

Highly accurate transcription results

Subtitle Generation

Automatically generate English subtitles for video content

Training Loss	Epoch	Step	Validation Loss	Wer
0.1131	1.12	1000	0.0755	0.0487
0.0725	2.24	2000	0.0637	0.0404
0.0539	3.36	3000	0.0661	0.0389
0.0441	4.48	4000	0.0637	0.0371
0.0379	5.61	5000	0.0675	0.0356
0.0341	6.73	6000	0.0735	0.0360
0.0295	7.85	7000	0.0737	0.0362
0.0265	8.97	8000	0.0741	0.0350
0.0244	10.09	9000	0.0779	0.0337
0.0217	11.21	10000	0.0835	0.0343
0.0203	12.33	11000	0.0785	0.0339
0.0188	13.45	12000	0.0827	0.0344
0.0179	14.57	13000	0.0875	0.0332
0.0169	15.7	14000	0.0860	0.0330
0.0158	16.82	15000	0.0954	0.0330
0.0147	17.94	16000	0.0934	0.0329
0.0148	19.06	17000	0.0965	0.0330

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

Librispeech 100h Supervised Meta

Model Overview

Model Features

Model Capabilities

Use Cases

🚀 librispeech-100h-supervised-meta

🚀 Quick Start

📚 Documentation

Model description

Intended uses & limitations

Training and evaluation data

🔧 Technical Details

Training procedure

Training hyperparameters

Training results

Framework versions

📄 License