IndicWav2Vec-Hindi
This is a Wav2Vec2 style ASR model trained in fairseq and ported to Hugging Face.
More details on datasets, training-setup and conversion to HuggingFace format can be found in the IndicWav2Vec repo.
Note: This model doesn't support inference with Language Model.
Script to Run Inference
import torch
from datasets import load_dataset
from transformers import AutoModelForCTC, AutoProcessor
import torchaudio.functional as F
DEVICE_ID = "cuda" if torch.cuda.is_available() else "cpu"
MODEL_ID = "ai4bharat/indicwav2vec-hindi"
sample = next(iter(load_dataset("common_voice", "hi", split="test", streaming=True)))
resampled_audio = F.resample(torch.tensor(sample["audio"]["array"]), 48000, 16000).numpy()
model = AutoModelForCTC.from_pretrained(MODEL_ID).to(DEVICE_ID)
processor = AutoProcessor.from_pretrained(MODEL_ID)
input_values = processor(resampled_audio, return_tensors="pt").input_values
with torch.no_grad():
logits = model(input_values.to(DEVICE_ID)).logits.cpu()
prediction_ids = torch.argmax(logits, dim=-1)
output_str = processor.batch_decode(prediction_ids)[0]
print(f"Greedy Decoding: {output_str}")
About AI4Bharat
- Website: https://ai4bharat.org/
- Code: https://github.com/AI4Bharat
- HuggingFace: https://huggingface.co/ai4bharat
- Downloads last month
- 1,277
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.