Open-source Stutter Detection Model distilhubert-finetuned-stutterdetection - Achieves 90.24% Accuracy on Evaluation Set

Distilhubert Finetuned Stutterdetection

Developed by HareemFatima

A stuttering detection model fine-tuned based on the DistilHuBERT model, achieving 90.24% accuracy on the evaluation set

Audio Classification

Transformers

Open Source License:Apache-2.0 #Speech disorder detection #High-precision classification #Medical auxiliary diagnosis

Downloads 116

Release Time : 4/27/2024

Model Overview

This model is a lightweight audio classification model based on the DistilHuBERT architecture, specifically designed to detect stuttering in speech. Suitable for speech pathology research or auxiliary diagnostic tool development.

Model Features

High accuracy

Achieves 90.24% classification accuracy on standard test sets

Lightweight architecture

Based on the distilled version of HuBERT, reducing computational resource requirements while maintaining performance

End-to-end training

Learns features directly from raw audio without manual feature design

Model Capabilities

Audio classification

Stuttering detection

Speech anomaly recognition

Use Cases

Medical assistance

Stuttering screening tool

Used for preliminary stuttering screening in clinical settings

Automatically identifies potential stuttering symptoms

Speech research

Speech pathology research

Assists researchers in analyzing stuttering characteristics

Provides objective quantitative metrics

🚀 distilhubert-finetuned-stutteringdetection

This model is a fine - tuned version of [ntu - spml/distilhubert](https://huggingface.co/ntu - spml/distilhubert) on the stuttering dataset. It can be used for audio classification tasks, achieving high accuracy on the evaluation set.

🚀 Quick Start

This model is a fine - tuned version of [ntu - spml/distilhubert](https://huggingface.co/ntu - spml/distilhubert) on the stuttering dataset. It achieves the following results on the evaluation set:

Loss: 0.5717
Accuracy: 0.9024

📚 Documentation

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e - 05
train_batch_size: 8
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 10
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8357	1.0	92	0.7812	0.8659
0.2951	2.0	184	0.3680	0.8902
0.097	3.0	276	0.4000	0.8659
0.0872	4.0	368	0.3953	0.9024
0.4557	5.0	460	0.4904	0.9024
0.0368	6.0	552	0.4972	0.9024
0.0074	7.0	644	0.5408	0.9146
0.0039	8.0	736	0.5460	0.9024
0.0036	9.0	828	0.5684	0.9024
0.0035	10.0	920	0.5717	0.9024

Framework versions

Transformers 4.40.1
Pytorch 2.2.1+cu121
Datasets 2.19.0
Tokenizers 0.19.1

📄 License

This model is licensed under the Apache 2.0 license.

📊 Model Information

Property	Details
Base Model	ntu - spml/distilhubert
Tags	generated_from_trainer
Datasets	HareemFatima/stutteringdetection
Metrics	accuracy
Model Name	distilhubert - finetuned - stutteringdetection
Task	Audio Classification
Dataset Name	stuttering
Dataset Type	HareemFatima/stutteringdetection
Metric Name	Accuracy
Metric Type	accuracy
Metric Value	0.9024390243902439

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