Neunit-ks Open-source Speech Processing Model - Achieving High-accuracy Speech Processing Based on Fine-tuning Technology

Neunit Ks

Developed by SHENMU007

A speech processing model fine-tuned based on facebook/wav2vec2-base, with an accuracy of 28.57%

Downloads 23

Release Time : 5/8/2023

Model Overview

This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset, primarily used for speech-related tasks

Fine-tuned based on wav2vec2-base

Uses the mature wav2vec2-base architecture as the base model for fine-tuning

Linear learning rate scheduling

Adopts a linear learning rate scheduling strategy during training, including a 10% warm-up phase

Speech processing

Speech feature extraction

Speech recognition

Basic speech recognition

Can be used for basic speech-to-text tasks

Accuracy 28.57%

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	25.0	25	0.7054	0.3214
No log	50.0	50	0.7030	0.3214
No log	75.0	75	0.7025	0.2857
No log	100.0	100	0.7024	0.2857

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base