Wav2vec2 Large Xlsr Hu

🚀 Wav2Vec2-Large-XLSR-53-Hungarian

This model is fine - tuned from facebook/wav2vec2-large-xlsr-53 on Hungarian using the Common Voice dataset. It's designed for speech recognition tasks, offering a solution for Hungarian language processing.

🚀 Quick Start

When using this model, make sure that your speech input is sampled at 16kHz.

✨ Features

• Fine - Tuned: Based on the large - scale pre - trained model facebook/wav2vec2-large-xlsr-53 and fine - tuned on the Hungarian language.
• Data Source: Trained on the Common Voice dataset, which provides a rich source of real - world speech data.

📦 Installation

No specific installation steps are provided in the original document.

💻 Usage Examples

Basic Usage

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("common_voice", "hu", split="test[:2%]")

processor = Wav2Vec2Processor.from_pretrained("gchhablani/wav2vec2-large-xlsr-hu")
model = Wav2Vec2ForCTC.from_pretrained("gchhablani/wav2vec2-large-xlsr-hu")

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["speech"][:2], 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["sentence"][:2])

Advanced Usage

The model can be evaluated as follows on the Portuguese test data of Common Voice.

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

test_dataset = load_dataset("common_voice", "hu", split="test")
wer = load_metric("wer")

processor = Wav2Vec2Processor.from_pretrained("gchhablani/wav2vec2-large-xlsr-hu")
model = Wav2Vec2ForCTC.from_pretrained("gchhablani/wav2vec2-large-xlsr-hu")
model.to("cuda")

chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\‘\”\�\–\…]'
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):
    batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
    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)

# Preprocessing the datasets.
# We need to read the aduio files as arrays
def evaluate(batch):
    inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

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

        pred_ids = torch.argmax(logits, dim=-1)
        batch["pred_strings"] = processor.batch_decode(pred_ids)
    return batch

result = test_dataset.map(evaluate, batched=True, batch_size=8)

print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))

Test Result: 46.75 %

📚 Documentation

Training

The Common Voice train and validation datasets were used for training. The code can be found here. The notebook containing the code used for evaluation can be found here.

📄 License

This model is licensed under the apache - 2.0 license.

Additional Information