Wav2vec2-3 Open-source Speech Recognition Model - Free to Use, Precise Speech Recognition with Low Error Rate

Wav2vec2 3

Developed by chrisvinsen

A fine-tuned speech recognition model based on facebook/wav2vec2-base with a Word Error Rate (WER) of 1.0

Downloads 16

Release Time : 5/22/2022

Model Overview

This model is a speech recognition model based on the wav2vec2 architecture, fine-tuned for speech-to-text tasks

Low Word Error Rate

Achieved a Word Error Rate (WER) of 1.0 on the evaluation set

Based on wav2vec2 Architecture

Fine-tuned using facebook/wav2vec2-base as the base model

Speech Recognition

Audio-to-Text Conversion

Speech Transcription

Meeting Minutes

Convert meeting recordings into text transcripts

Voice Notes

Convert voice memos into text

This model is a fine - tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

This section provides an overview of the wav2vec2 - 3 model and its performance on the evaluation set.

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

Training Loss	Epoch	Step	Validation Loss	Wer
3.7797	0.34	200	3.0703	1.0
2.8701	0.69	400	3.3128	1.0
2.8695	1.03	600	3.1333	1.0
2.8634	1.38	800	3.1634	1.0
2.8629	1.72	1000	3.0432	1.0
2.8652	2.07	1200	3.0300	1.0
2.8602	2.41	1400	3.1894	1.0
2.8622	2.75	1600	3.1950	1.0
2.8606	3.1	1800	3.0656	1.0
2.8605	3.44	2000	3.0614	1.0
2.8595	3.79	2200	3.0697	1.0
2.8504	4.13	2400	3.1404	1.0
2.8553	4.48	2600	3.0682	1.0
2.8585	4.82	2800	3.1393	1.0
2.8567	5.16	3000	3.1013	1.0
2.8539	5.51	3200	3.0740	1.0
2.8588	5.85	3400	3.0616	1.0
2.8509	6.2	3600	3.1032	1.0
2.8589	6.54	3800	3.1348	1.0
2.8505	6.88	4000	3.1514	1.0
2.8548	7.23	4200	3.1319	1.0
2.8466	7.57	4400	3.1412	1.0
2.8549	7.92	4600	3.1235	1.0
2.8532	8.26	4800	3.0751	1.0
2.8548	8.61	5000	3.0946	1.0
2.8513	8.95	5200	3.0840	1.0
2.845	9.29	5400	3.0896	1.0
2.8592	9.64	5600	3.1055	1.0
2.8453	9.98	5800	3.1124	1.0