Create README.md

README.md

@@ -0,0 +1,50 @@
+---
+language: or
+metrics:
+- wer
+- cer
+tags:
+- audio
+- automatic-speech-recognition
+- speech
+- wav2vec2
+- asr
+license: apache-2.0
+---
+
+# IndicWav2Vec-Hindi
+
+This is a [Wav2Vec2](https://arxiv.org/abs/2006.11477) style ASR model trained in [fairseq](https://github.com/facebookresearch/fairseq) and ported to Hugging Face. 
+More details on datasets, training-setup and conversion to HuggingFace format can be found in the [IndicWav2Vec](https://github.com/AI4Bharat/IndicWav2Vec) repo.  
+
+## Script to Run Inference
+
+```python
+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-odia"
+
+sample = next(iter(load_dataset("common_voice", "or", 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