Downloads · 30 days
9
17% of all-time downloads
costin-b/wav2vec2-base-100k-voxpopuli-romanian
wav2vec2-base-100k-voxpopuli-romanian is a machine learning model from costin-b. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as mit.
The model was fine-tuned on 300h of public and private speech data. More information will be given once the underlying paper gets published.
Downloads · 30 days
9
17% of all-time downloads
All-time downloads
54
Public
Repo size
755 MB
Likes
0
Public
Click a slice to open those files.
.bin378 MB · 100%
From the Hugging Face model README
The model was fine-tuned on 300h of public and private speech data. More information will be given once the underlying paper gets published.
import librosa
from transformers import Wav2Vec2Processor, AutoModelForCTC
import torch
audio, _ = librosa.load("[audio_path]", sr=16000)
model = AutoModelForCTC.from_pretrained("racai/wav2vec2-base-100k-voxpopuli-romanian")
processor = Wav2Vec2Processor.from_pretrained("racai/wav2vec2-base-100k-voxpopuli-romanian")
input_dict = processor(audio, sampling_rate=16000, return_tensors="pt")
with torch.inference_mode():
logits = model(input_dict.input_values).logits
predicted_ids = torch.argmax(logits, dim=-1)
predicted_sentence = processor.batch_decode(predicted_ids)[0]
print("Prediction:", predicted_sentence)