đ Fine-tuned XLS-R 1B model for speech recognition in Dutch
This is a fine-tuned model for Dutch speech recognition. It is based on facebook/wav2vec2-xls-r-1b, fine-tuned on Dutch using the train and validation splits of Common Voice 8.0, Multilingual LibriSpeech, and Voxpopuli. When using this model, ensure that your speech input is sampled at 16kHz.
đ Quick Start
This model has been fine-tuned by the HuggingSound tool, thanks to the GPU credits generously provided by the OVHcloud.
⨠Features
- Automatic Speech Recognition: Capable of accurately transcribing Dutch speech.
- Fine-tuned on Multiple Datasets: Trained on Common Voice 8.0, Multilingual LibriSpeech, and Voxpopuli for better performance.
đĻ Installation
There is no specific installation command provided in the original document, so this section is skipped.
đģ Usage Examples
Basic Usage
Using the HuggingSound library:
from huggingsound import SpeechRecognitionModel
model = SpeechRecognitionModel("jonatasgrosman/wav2vec2-xls-r-1b-dutch")
audio_paths = ["/path/to/file.mp3", "/path/to/another_file.wav"]
transcriptions = model.transcribe(audio_paths)
Advanced Usage
Writing your own inference script:
import torch
import librosa
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
LANG_ID = "nl"
MODEL_ID = "jonatasgrosman/wav2vec2-xls-r-1b-dutch"
SAMPLES = 10
test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")
processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
def speech_file_to_array_fn(batch):
speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
batch["speech"] = speech_array
batch["sentence"] = batch["sentence"].upper()
return batch
test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["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)
predicted_sentences = processor.batch_decode(predicted_ids)
đ Documentation
Evaluation Commands
- To evaluate on
mozilla-foundation/common_voice_8_0
with split test
python eval.py --model_id jonatasgrosman/wav2vec2-xls-r-1b-dutch --dataset mozilla-foundation/common_voice_8_0 --config nl --split test
- To evaluate on
speech-recognition-community-v2/dev_data
python eval.py --model_id jonatasgrosman/wav2vec2-xls-r-1b-dutch --dataset speech-recognition-community-v2/dev_data --config nl --split validation --chunk_length_s 5.0 --stride_length_s 1.0
Model Information
Property |
Details |
Model Type |
Fine-tuned XLS-R 1B for Dutch speech recognition |
Training Data |
mozilla-foundation/common_voice_8_0, Multilingual LibriSpeech, Voxpopuli |
Results
Task |
Dataset |
Test WER |
Test CER |
Test WER (+LM) |
Test CER (+LM) |
Dev WER |
Dev CER |
Dev WER (+LM) |
Dev CER (+LM) |
Automatic Speech Recognition |
Common Voice 8 |
10.38 |
3.04 |
6.83 |
2.31 |
- |
- |
- |
- |
Automatic Speech Recognition |
Robust Speech Event - Dev Data |
- |
- |
- |
- |
31.12 |
15.92 |
23.95 |
14.18 |
Automatic Speech Recognition |
Robust Speech Event - Test Data |
20.41 |
- |
- |
- |
- |
- |
- |
- |
đ License
This model is licensed under the Apache-2.0 license.
đ Citation
If you want to cite this model you can use this:
@misc{grosman2021xlsr-1b-dutch,
title={Fine-tuned {XLS-R} 1{B} model for speech recognition in {D}utch},
author={Grosman, Jonatas},
howpublished={\url{https://huggingface.co/jonatasgrosman/wav2vec2-xls-r-1b-dutch}},
year={2022}
}