AIbase
EN
Home
Wav2vec2 Large 960h Lv60 Self MIDIARIES 72H FT
W

Wav2vec2 Large 960h Lv60 Self MIDIARIES 72H FT

Developed by caurdy
A speech recognition model fine-tuned using 72 hours of MI diary data, based on Facebook's pre-trained wav2vec2 large 960H lv60 self-supervised model
Speech Recognition
Transformers
#Speech recognition optimization#Medical diary transcription#Fine-tuning error reduction
Downloads 20
Release Time : 4/21/2022

Model Overview

This model is specifically optimized for medical interview scenarios, significantly improving recognition accuracy in medical dialogue contexts through fine-tuning

Model Features

Medical scenario optimization
Fine-tuned with 72 hours of medical interview data, particularly suitable for medical dialogue scenarios
Performance improvement
On a 20-minute MI diary test set, the word error rate decreased from 13% to 9.7%
Based on mature architecture
Built upon Facebook's pre-trained wav2vec2 large 960H lv60 self-supervised model

Model Capabilities

English speech recognition
Medical dialogue transcription
Automatic speech-to-text conversion

Use Cases

Healthcare
Medical interview recording
Automatically transcribes conversations between doctors and patients
Word error rate reduced to 9.7%
Medical document generation
Automatically converts medical interview recordings into structured documents
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase