---
license: cc-by-nc-4.0
language:
- bn
library_name: nemo
pipeline_tag: automatic-speech-recognition
---
## Hishab BN FastConformer
__Hishab BN FastConformer__ is a [fastconformer](https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/stable/asr/models.html#fast-conformer) based model trained on ~18K Hours [MegaBNSpeech]() corpus.

## Using method
This model can be used for transcribing Bangla audio and also can be used as pre-trained model to fine-tuning on custom datasets using [NeMo](https://github.com/NVIDIA/NeMo) framework.

### Installation
To install [NeMo](https://github.com/NVIDIA/NeMo) check NeMo documentation.

### Inferencing
```py
import nemo.collections.asr as nemo_asr
asr_model = nemo_asr.models.ASRModel.from_pretrained("hishab/hishab_bn_fastconformer")

transcriptions = asr_model.transcribe(["file.wav"])
```
## Training Datasets

| Channels Category | Hours       |
| ----------------- | ----------- |
| News             | 17,640.00   |
| Talkshow         | 688.82      |
| Vlog             | 0.02        |
| Crime Show       | 4.08        |
| Total            | 18,332.92   |


## Training Details

For training the model, the dataset we selected comprises 17.64k hours of news chan- nel content, 688.82 hours of talk shows, 0.02 hours of vlogs, and 4.08 hours of crime shows.

## Evaluation


![image/png](https://cdn-uploads.huggingface.co/production/uploads/64df9253cccd823564c3303b/WvMlp95z2-GXT6AYfwW8Y.png)

![image/png](https://cdn-uploads.huggingface.co/production/uploads/64df9253cccd823564c3303b/O2RA9TAedIv1OTqgdIap5.png)

## Citation