Finetune_indian_asr Open-source Model - Accurately Recognize Indian English Speech, Free and Practical!

Home

Finetune Indian Asr

Developed by Simply-divine

Indian English speech recognition model fine-tuned based on Harveenchadha/vakyansh-wav2vec2-indian-english-enm-700

Speech Recognition

Transformers

#Indian English Speech Recognition #Low Word Error Rate #Wav2Vec2 Fine-tuning

Downloads 20

Release Time : 3/14/2022

Model Overview

This model is an automatic speech recognition (ASR) model optimized for Indian English, significantly reducing the word error rate (WER) after fine-tuning.

Model Features

Low Word Error Rate

After fine-tuning, the validation set word error rate (WER) was reduced to 0.3403

Indian English Optimization

Specifically optimized for Indian English datasets, suitable for Indian accent recognition

Efficient Training

Using linear learning rate scheduling and Adam optimizer, 30 training rounds achieved good results

Model Capabilities

Indian English Speech-to-Text

Continuous Speech Recognition

Accent-Adaptive Recognition

Use Cases

Speech Transcription

Customer Service Call Transcription

Automatically transcribe Indian English customer service call content

Word error rate 0.34

Meeting Minutes

Real-time transcription of Indian English meeting content

Training Loss	Epoch	Step	Validation Loss	Wer
5.0566	3.45	500	2.9944	1.0
2.7241	6.9	1000	1.4455	0.7654
0.9755	10.34	1500	0.4299	0.4034
0.4624	13.79	2000	0.3628	0.3297
0.3158	17.24	2500	0.3835	0.2952
0.2604	20.69	3000	0.3802	0.2877
0.2	24.14	3500	0.3842	0.2799
1.7441	27.59	4000	0.4215	0.3403

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

Finetune Indian Asr

Model Overview

Model Features

Model Capabilities

Use Cases

🚀 finetune_indian_asr

🚀 Quick Start

📚 Documentation

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions