wav2vec2-base-timit-demo-colab-1 Open-source Speech Recognition Model - Accurately Identify Speech Content with Low Error Rate

Wav2vec2 Base Timit Demo Colab 1

Developed by zasheza

This model is a fine-tuned speech recognition model based on facebook/wav2vec2-base, trained on the TIMIT dataset with a word error rate (WER) of 0.4398.

Speech Recognition

Transformers

Open Source License:Apache-2.0 #Speech Recognition #Low Word Error Rate #TIMIT Dataset

Downloads 18

Release Time : 5/1/2022

Model Overview

A speech recognition model based on the wav2vec2 architecture, suitable for English speech-to-text tasks.

Model Features

Based on wav2vec2 Architecture

Utilizes the open-source wav2vec2-base model architecture from Facebook, which has excellent speech feature extraction capabilities.

Fine-tuned Optimization

Fine-tuned on the TIMIT dataset for optimized performance on specific speech recognition tasks.

Relatively Low Word Error Rate

Achieves a word error rate (WER) of 0.4398 on the evaluation set, outperforming the base model.

Model Capabilities

English Speech Recognition

Speech-to-Text

Use Cases

Speech Transcription

Meeting Minutes

Automatically transcribe English meeting recordings into text

Accuracy approximately 56.02% (1-WER)

Voice Notes

Convert English voice notes into searchable text

Training Loss	Epoch	Step	Validation Loss	Wer
4.8991	5.26	500	1.4319	0.7522
0.8555	10.53	1000	0.7895	0.5818
0.4584	15.79	1500	0.7198	0.5211
0.3096	21.05	2000	0.7983	0.5118
0.2165	26.32	2500	0.7893	0.4745
0.163	31.58	3000	0.8779	0.4589
0.1144	36.84	3500	0.9256	0.4540
0.0886	42.11	4000	0.9184	0.4530
0.0668	47.37	4500	0.9634	0.4398

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

Wav2vec2 Base Timit Demo Colab 1

Model Overview

Model Features

Model Capabilities

Use Cases

🚀 wav2vec2-base-timit-demo-colab-1

🚀 Quick Start

📚 Documentation

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

📄 License