romanian-wav2vec2

You can test this model online with the [**Space for Romanian Speech Recognition**](https://huggingface.co/spaces/gigant/romanian-speech-recognition) The model ranked **TOP-1** on Romanian Speech Recognition during HuggingFace's Robust Speech Challenge : * [**The 🤗 Speech Bench**](https://huggingface.co/spaces/huggingface/hf-speech-bench) * [**Speech Challenge Leaderboard**](https://huggingface.co/spaces/speech-recognition-community-v2/FinalLeaderboard) This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the [Common Voice 8.0 - Romanian subset](https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0) dataset, with extra training data from [Romanian Speech Synthesis](https://huggingface.co/datasets/gigant/romanian_speech_synthesis_0_8_1) dataset. Without the 5-gram Language Model optimization, it achieves the following results on the evaluation set (Common Voice 8.0, Romanian subset, test split): - Loss: 0.1553 - Wer: 0.1174 - Cer: 0.0294 The architecture is based on [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) with a speech recognition CTC head and an added 5-gram language model (using [pyctcdecode](https://github.com/kensho-technologies/pyctcdecode) and [kenlm](https://github.com/kpu/kenlm)) trained on the [Romanian Corpora Parliament](gigant/ro_corpora_parliament_processed) dataset. Those libraries are needed in order for the language model-boosted decoder to work. The model is made for speech recognition in Romanian from audio clips sampled at **16kHz**. The predicted text is lowercased and does not contain any punctuation. Make sure you have installed the correct dependencies for the language model-boosted version to work. You can just run this command to install the `kenlm` and `pyctcdecode` libraries : With the framework `transformers` you can load the model with the following code : from transformers import AutoProcessor, AutoModelForCTC processor = AutoProcessor.from_pretrained("gigant/romanian-wav2vec2") model = AutoModelForCTC.from_pretrained("gigant/romanian-wav2vec2") Or, if you want to test the model, you can load the automatic speech recognition pipeline from `transformers` with : from transformers import pipeline asr = pipeline("automatic-speech-recognition", model="gigant/romanian-wav2vec2") First, you need to load your data We will use the [Romanian Speech Synthesis](https://huggingface.co/datasets/gigant/romanian_speech_synthesis_0_8_1) dataset in this example. from datasets import load_dataset dataset = load_dataset("gigant/romanian_speech_synthesis_0_8_1") You can listen to the samples with the `IPython.display` library : from IPython.display import Audio i = 0 sample = dataset["train"][i] Audio(sample["audio"]["array"], rate = sample["audio"]["sampling_rate"]) The model is trained to work with audio sampled at 16kHz, so if the sampling rate of the audio in the dataset is different, we will have to resample it. In the example, the audio is sampled at 48kHz. We can see this by checking `dataset["train"][0]["audio"]["sampling_rate"]` The following code resample the audio using the `torchaudio` library : import torchaudio import torch i = 0 audio = sample["audio"]["array"] rate = sample["audio"]["sampling_rate"] resampler = torchaudio.transforms.Resample(rate, 16_000) audio_16 = resampler(torch.Tensor(audio)).numpy() To listen to the resampled sample : Audio(audio_16, rate=16000) Know you can get the model prediction by running predicted_text = asr(audio_16) ground_truth = dataset["train"][i]["sentence"] print(f"Predicted text : {predicted_text}") print(f"Ground truth : {ground_truth}") Training data : - [Common Voice 8.0 - Romanian subset](https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0) : train + validation + other splits - [Romanian Speech Synthesis](https://huggingface.co/datasets/gigant/romanian_speech_synthesis_0_8_1) : train + test splits Evaluation data : - [Common Voice 8.0 - Romanian subset](https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0) : test split The following hyperparameters were used during training: - learning_rate: 0.003 - train_batch_size: 16 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 3 - total_train_batch_size: 48 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 500 - num_epochs: 50.0 - mixed_precision_training: Native AMP - Transformers 4.16.2 - Pytorch 1.10.0+cu111 - Tokenizers 0.11.0 - pyctcdecode 0.3.0 - kenlm