wav2vec2-tiny-random-rodela-classifier

This model is a fine-tuned version of patrickvonplaten/wav2vec2_tiny_random on shhossain/rodela_dataset dataset. It achieves the following results on the evaluation set:

• eval_loss: 0.7650
• eval_accuracy: 0.9102
• eval_runtime: 0.2323
• eval_samples_per_second: 1054.546
• eval_steps_per_second: 133.432
• epoch: 37.0
• step: 4551

Model description

It classifies if a audio has rodela in it.

Intended uses & limitations

It was designed for Wake Word Detection.

How to use

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("audio-classification", model="shhossain/wav2vec2-tiny-random-rodela-classifier")
pipe("my_audio.mp3")

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• total_train_batch_size: 4.0
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100
• gradient_accumulation_steps: .5

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2