Wynehills-mimi-ASR Open-Source Automatic Speech Recognition Model - Accurately Recognize Speech Content with Low Error Rate

Wynehills Mimi ASR

Developed by mimi

This is an Automatic Speech Recognition (ASR) model trained on an unknown dataset, achieving a Word Error Rate (WER) of 0.6309.

Speech Recognition

Transformers

#Automatic Speech Recognition #Low Word Error Rate #Linear Learning Rate Scheduling

Downloads 26

Release Time : 3/2/2022

Model Overview

This model is an automatic speech recognition system capable of converting speech into text. The training utilized the Adam optimizer and a linear learning rate scheduler.

Model Features

Optimized Training Process

Trained using the Adam optimizer and a linear learning rate scheduler with a learning rate of 0.0001.

Mixed Precision Training

Utilizes native AMP (Automatic Mixed Precision) training technology to enhance training efficiency.

Model Capabilities

Speech-to-Text

Automatic Speech Recognition

Use Cases

Speech Transcription

Meeting Minutes

Automatically convert meeting recordings into text transcripts.

Word Error Rate 0.6309

Voice Memo Transcription

Convert voice memos into editable text.

🚀 wynehills-mimi-ASR

This is an Automatic Speech Recognition (ASR) model trained from scratch. It was trained on an unknown dataset and achieves a loss of 1.3822 and a Word Error Rate (WER) of 0.6309 on the evaluation set.

📚 Documentation

Model Performance

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 1.3822
Wer: 0.6309

Training Procedure

Training Hyperparameters

The following hyperparameters were used during training:

Property	Details
learning_rate	0.0001
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_steps	1000
num_epochs	70
mixed_precision_training	Native AMP

Training Results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.54	20	1.4018	0.6435
No log	3.08	40	1.4704	0.6593
No log	4.62	60	1.4898	0.6625
No log	6.15	80	1.4560	0.6404
No log	7.69	100	1.3822	0.6309
No log	9.23	120	1.3822	0.6309
No log	10.77	140	1.3822	0.6309
No log	12.31	160	1.3822	0.6309
No log	13.85	180	1.3822	0.6309
No log	15.38	200	1.3822	0.6309
No log	16.92	220	1.3822	0.6309
No log	18.46	240	1.3822	0.6309
No log	20.0	260	1.3822	0.6309
No log	21.54	280	1.3822	0.6309
No log	23.08	300	1.3822	0.6309
No log	24.62	320	1.3822	0.6309
No log	26.15	340	1.3822	0.6309
No log	27.69	360	1.3822	0.6309
No log	29.23	380	1.3822	0.6309
No log	30.77	400	1.3822	0.6309
No log	32.31	420	1.3822	0.6309
No log	33.85	440	1.3822	0.6309
No log	35.38	460	1.3822	0.6309
No log	36.92	480	1.3822	0.6309
0.0918	38.46	500	1.3822	0.6309
0.0918	40.0	520	1.3822	0.6309
0.0918	41.54	540	1.3822	0.6309
0.0918	43.08	560	1.3822	0.6309
0.0918	44.62	580	1.3822	0.6309
0.0918	46.15	600	1.3822	0.6309
0.0918	47.69	620	1.3822	0.6309
0.0918	49.23	640	1.3822	0.6309
0.0918	50.77	660	1.3822	0.6309
0.0918	52.31	680	1.3822	0.6309
0.0918	53.85	700	1.3822	0.6309
0.0918	55.38	720	1.3822	0.6309
0.0918	56.92	740	1.3822	0.6309
0.0918	58.46	760	1.3822	0.6309
0.0918	60.0	780	1.3822	0.6309
0.0918	61.54	800	1.3822	0.6309
0.0918	63.08	820	1.3822	0.6309
0.0918	64.62	840	1.3822	0.6309
0.0918	66.15	860	1.3822	0.6309
0.0918	67.69	880	1.3822	0.6309
0.0918	69.23	900	1.3822	0.6309

Framework Versions

Transformers 4.11.3
Pytorch 1.10.0+cu111
Datasets 1.13.3
Tokenizers 0.10.3

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