Whisper-medium-tr Open-source Speech Recognition Model - Free Deployment, Significantly Reduce the Error Rate of Turkish Speech Recognition

Whisper Medium Tr

Developed by sgangireddy

This model is a fine-tuned Turkish speech recognition model based on OpenAI's Whisper Medium, with significantly reduced word error rate.

Speech Recognition

Transformers

OtherOpen Source License:Apache-2.0 #Turkish speech recognition #Low word error rate #Multi-scenario applicability

Downloads 81

Release Time : 12/7/2022

Model Overview

Whisper Medium Turkish is an automatic speech recognition (ASR) model optimized for Turkish, fine-tuned on the Common Voice 11.0 Turkish dataset, significantly improving recognition accuracy.

Model Features

High-performance Turkish recognition

Achieves a word error rate of only 10.50% on the Common Voice 11.0 test set, a 49% improvement over zero-shot performance

Multi-scenario adaptability

Performs excellently on the Google Fluers dataset as well, with a word error rate of 9.26%

Efficient fine-tuning

Significant performance improvement achieved with just 1000 training steps

Model Capabilities

Turkish speech recognition

Speech-to-text

Multi-scenario speech processing

Use Cases

Speech transcription

Turkish speech-to-text

Converts Turkish speech content into accurate text

Word error rate 10.50%

Voice assistants

Turkish voice interaction

Provides core recognition capabilities for Turkish voice assistants

🚀 Whisper Medium Turkish

This model is a fine - tuned version of [openai/whisper - medium](https://huggingface.co/openai/whisper - medium) on the mozilla - foundation/common_voice_11_0 Turkish dataset. It can effectively perform automatic speech recognition tasks on Turkish audio, providing high - quality transcription results.

🚀 Quick Start

This model is a fine - tuned version of [openai/whisper - medium](https://huggingface.co/openai/whisper - medium) on the mozilla - foundation/common_voice_11_0 Turkish dataset. It achieves the following results on the evaluation set:

Loss: 0.1879
Wer: 10.5033

✨ Features

Fine - Tuned on Turkish Dataset: Trained on the mozilla - foundation/common_voice_11_0 Turkish dataset, it is well - adapted to Turkish speech characteristics.
Performance Improvement: Compared with the zero - shot model, it shows significant improvement in word error rate (Wer).

📚 Documentation

Model description

The model is fine - tuned for 1000 steps/updates.

Zero - shot - 20.89 (CV11)
Fine - tune on CV11 - 10.50 (CV11) (-49%)

Zeroshot - 10.4 (Google Fluers)
Fine - tune on CV11 - 9.26 (Google Fluers)

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: 1e - 05
train_batch_size: 32
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 64
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 1000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0348	3.05	1000	0.1879	10.5033

Framework versions

Transformers 4.26.0.dev0
Pytorch 1.13.0+cu117
Datasets 2.7.1.dev0
Tokenizers 0.13.2

📄 License

This model is licensed under the Apache - 2.0 license.

Property	Details
Model Type	Fine - tuned version of [openai/whisper - medium](https://huggingface.co/openai/whisper - medium) on Turkish dataset
Training Data	mozilla - foundation/common_voice_11_0 Turkish dataset

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