wav2vec2_base_10k_8khz_pt_cv7_2 Open Source Model - Free Deployment for 8kHz Portuguese Speech Recognition

Wav2vec2 Base 10k 8khz Pt Cv7 2

Developed by lgris

This model is a Portuguese automatic speech recognition model based on the wav2vec2 architecture, fine-tuned on the Common Voice 7 dataset, supporting 8kHz sample rate audio input.

Speech Recognition

Transformers

OtherOpen Source License:Apache-2.0 #Portuguese speech recognition #8kHz audio optimization #Common Voice fine-tuning

Downloads 24

Release Time : 3/2/2022

Model Overview

A Portuguese speech recognition model based on the wav2vec2 architecture, optimized for 8kHz sample rate audio, fine-tuned on the Common Voice 7 dataset, suitable for Portuguese speech-to-text tasks.

Model Features

Portuguese Optimization

Specifically optimized and fine-tuned for Portuguese speech recognition tasks

8kHz Sample Rate Support

Supports 8kHz sample rate audio input, suitable for telephone-quality speech recognition

Common Voice 7 Fine-tuning

Fine-tuned on the Mozilla Common Voice 7 Portuguese dataset

Model Capabilities

Portuguese speech recognition

8kHz audio processing

Automatic speech-to-text

Use Cases

Speech Transcription

Portuguese Speech Transcription

Convert Portuguese speech content into text

WER of 36.9 on the Common Voice 7 test set

Voice Assistants

Portuguese Voice Command Recognition

Recognize and understand Portuguese voice commands

🚀 wav2vec2_base_10k_8khz_pt_cv7_2

This model is a fine - tuned version of lgris/seasr_2022_base_10k_8khz_pt on the common_voice dataset, achieving certain results on the evaluation set.

🚀 Quick Start

This model is a fine - tuned version of lgris/seasr_2022_base_10k_8khz_pt on the common_voice dataset. It achieves the following results on the evaluation set:

Loss: 76.3426
Wer: 0.1979

📚 Documentation

Model Information

Property	Details
Model Type	Fine - tuned version of lgris/seasr_2022_base_10k_8khz_pt on the common_voice dataset
Training Data	mozilla - foundation/common_voice_7_0
Tags	automatic - speech - recognition, generated_from_trainer, hf - asr - leaderboard, mozilla - foundation/common_voice_7_0, pt, robust - speech - event

Results on Datasets

Task	Dataset	Metrics	Value
Automatic Speech Recognition	Common Voice 7 (pt)	Test WER	36.9
Automatic Speech Recognition	Common Voice 7 (pt)	Test CER	14.82
Automatic Speech Recognition	Robust Speech Event - Dev Data (sv)	Test WER	40.53
Automatic Speech Recognition	Robust Speech Event - Dev Data (sv)	Test CER	16.95
Automatic Speech Recognition	Robust Speech Event - Dev Data (pt)	Test WER	37.15
Automatic Speech Recognition	Robust Speech Event - Test Data (pt)	Test WER	38.95

🔧 Technical Details

Training Hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e - 05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 100
training_steps: 10000

Training Results

Training Loss	Epoch	Step	Validation Loss	Wer
189.1362	0.65	500	80.6347	0.2139
174.2587	1.3	1000	80.2062	0.2116
164.676	1.95	1500	78.2161	0.2073
176.5856	2.6	2000	78.8920	0.2074
164.3583	3.25	2500	77.2865	0.2066
161.414	3.9	3000	77.8888	0.2048
158.283	4.55	3500	77.3472	0.2033
159.2265	5.19	4000	79.0953	0.2036
156.3967	5.84	4500	76.6855	0.2029
154.2743	6.49	5000	77.7785	0.2015
156.6497	7.14	5500	77.1220	0.2033
157.3038	7.79	6000	76.2926	0.2027
162.8151	8.44	6500	76.7602	0.2013
151.8613	9.09	7000	77.4777	0.2011
153.0225	9.74	7500	76.5206	0.2001
157.52	10.39	8000	76.1061	0.2006
145.0592	11.04	8500	76.7855	0.1992
150.0066	11.69	9000	76.0058	0.1988
146.8128	12.34	9500	76.2853	0.1987
146.9148	12.99	10000	76.3426	0.1979

Framework Versions

Transformers 4.16.2
Pytorch 1.10.0+cu111
Datasets 1.18.3
Tokenizers 0.11.0

📄 License

This model is under the apache - 2.0 license.

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