---
language: vi
datasets:
- youtube-vi-13k-hours
tags:
- speech
license: cc-by-nc-4.0
---

# Vietnamese Self-Supervised Learning Wav2Vec2 model

## Model

We use wav2vec2 architecture for doing Self-Supervised learning

<img src="https://raw.githubusercontent.com/patrickvonplaten/scientific_images/master/wav2vec2.png" width=75% height=75%>

## Data

Our self-supervised model is pre-trained on a massive audio set of 13k hours of Vietnamese youtube audio, which includes:
  - Clean audio
  - Noise audio
  - Conversation
  - Multi-gender and dialects
  

## Download

We have already upload our pre-trained model to the Huggingface. The base model trained 35 epochs and the large model trained 20 epochs in about 30 days using TPU V3-8.

 - [Based version](https://huggingface.co/nguyenvulebinh/wav2vec2-base-vi) ~ 95M params
 - [Large version](https://huggingface.co/nguyenvulebinh/wav2vec2-large-vi) ~ 317M params

## Usage

```python
from transformers import Wav2Vec2ForPreTraining, Wav2Vec2Processor

model_name = 'nguyenvulebinh/wav2vec2-base-vi'
# model_name = 'nguyenvulebinh/wav2vec2-large-vi'

model = Wav2Vec2ForPreTraining.from_pretrained(model_name)
processor = Wav2Vec2Processor.from_pretrained(model_name)

```

Since our model has the same architecture as the English wav2vec2 version, you can use [this notebook](https://colab.research.google.com/drive/1FjTsqbYKphl9kL-eILgUc-bl4zVThL8F?usp=sharing) for more information on how to fine-tune the model.

## Finetuned version

### VLSP 2020 ASR dataset

Benchmark WER result on VLSP T1 testset:

| | [base model](https://huggingface.co/nguyenvulebinh/wav2vec2-base-vi-vlsp2020) | [large model](https://huggingface.co/nguyenvulebinh/wav2vec2-large-vi-vlsp2020) |
|---|---|---|
|without LM| 8.66  | 6.90 |
|with 5-grams LM| 6.53 | 5.32 |

Usage

```python
#pytorch
#!pip install transformers==4.20.0
#!pip install https://github.com/kpu/kenlm/archive/master.zip
#!pip install pyctcdecode==0.4.0
from transformers.file_utils import cached_path, hf_bucket_url
from importlib.machinery import SourceFileLoader
from transformers import Wav2Vec2ProcessorWithLM
from IPython.lib.display import Audio
import torchaudio
import torch

# Load model & processor
model_name = "nguyenvulebinh/wav2vec2-base-vi-vlsp2020"
# model_name = "nguyenvulebinh/wav2vec2-large-vi-vlsp2020"
model = SourceFileLoader("model", cached_path(hf_bucket_url(model_name,filename="model_handling.py"))).load_module().Wav2Vec2ForCTC.from_pretrained(model_name)
processor = Wav2Vec2ProcessorWithLM.from_pretrained(model_name)

# Load an example audio (16k)
audio, sample_rate = torchaudio.load(cached_path(hf_bucket_url(model_name, filename="t2_0000006682.wav")))
input_data = processor.feature_extractor(audio[0], sampling_rate=16000, return_tensors='pt')

# Infer
output = model(**input_data)

# Output transcript without LM
print(processor.tokenizer.decode(output.logits.argmax(dim=-1)[0].detach().cpu().numpy()))

# Output transcript with LM
print(processor.decode(output.logits.cpu().detach().numpy()[0], beam_width=100).text)
```

## Acknowledgment

- We would like to thank the Google TPU Research Cloud (TRC) program and Soonson Kwon (Google ML Ecosystem programs Lead) for their support.
- Special thanks to my colleagues at [VietAI](https://vietai.org/) and [VAIS](https://vais.vn/) for their advice.

## Contact 

nguyenvulebinh@gmail.com / binh@vietai.org

[![Follow](https://img.shields.io/twitter/follow/nguyenvulebinh?style=social)](https://twitter.com/intent/follow?screen_name=nguyenvulebinh)