wav2vec2-xls-r-300m-pl-cv8

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice 8.0 dataset. It achieves the following results on the evaluation set while training:

• Loss: 0.1716
• Wer: 0.1697
• Cer: 0.0385

The eval.py script results are: WER: 0.16970531733661967 CER: 0.03839135416519316

Model description

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Polish using the Common Voice dataset. When using this model, make sure that your speech input is sampled at 16kHz.

The model can be used directly (without a language model) as follows:

import torch
import torchaudio
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

test_dataset = load_dataset("mozilla-foundation/common_voice_8_0", "pl", split="test[:2%]")

processor = Wav2Vec2Processor.from_pretrained("comodoro/wav2vec2-xls-r-300m-pl-cv8")
model = Wav2Vec2ForCTC.from_pretrained("comodoro/wav2vec2-xls-r-300m-pl-cv8")

resampler = torchaudio.transforms.Resample(48_000, 16_000)

# Preprocessing the datasets.
# We need to read the aduio files as arrays
def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = torchaudio.load(batch["path"])
    batch["speech"] = resampler(speech_array).squeeze().numpy()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset[:2]["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

with torch.no_grad():
    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

predicted_ids = torch.argmax(logits, dim=-1)

print("Prediction:", processor.batch_decode(predicted_ids))
print("Reference:", test_dataset[:2]["sentence"])

Evaluation

The model can be evaluated using the attached eval.py script:

python eval.py --model_id comodoro/wav2vec2-xls-r-300m-pl-cv8 --dataset mozilla-foundation/common-voice_8_0 --split test --config pl

Training and evaluation data

The Common Voice 8.0 train and validation datasets were used for training

Training procedure

Training hyperparameters

The following hyperparameters were used:

• learning_rate: 1e-4
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 1
• total_train_batch_size: 640
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 150
• mixed_precision_training: Native AMP

The training was interrupted after 3250 steps.

Framework versions

• Transformers 4.16.0.dev0
• Pytorch 1.10.1+cu102
• Datasets 1.17.1.dev0
• Tokenizers 0.11.0