Indicconformer Stt Ne Hybrid Ctc Rnnt Large

🚀 IndicConformer

IndicConformer is a Hybrid CTC - RNNT conformer ASR (Automatic Speech Recognition) model, designed to transcribe Nepali speech.

🚀 Quick Start

To start using the IndicConformer model, you first need to install AI4Bharat NeMo. We recommend using the following command for installation:

git clone https://github.com/AI4Bharat/NeMo.git && cd NeMo && git checkout nemo-v2 && bash reinstall.sh

✨ Features

Language

This model is trained for Nepali language processing.

Input

It accepts 16000 KHz Mono - channel Audio (wav files) as input.

Output

Given an audio sample, the model provides the transcribed speech as a string.

🔧 Technical Details

Model Architecture

This is a conformer - Large model. The encoder has 120M parameters and is paired with a hybrid CTC - RNNT decoder. The model contains 17 conformer blocks, with a model dimension of 512.

💻 Usage Examples

Basic Usage

First, download and load the model from Huggingface:

import torch
import nemo.collections.asr as nemo_asr

model = nemo_asr.models.ASRModel.from_pretrained("ai4bharat/indicconformer_stt_ne_hybrid_rnnt_large")

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.freeze() # inference mode
model = model.to(device) # transfer model to device

Prepare an audio file by converting it to 16000 Hz and monochannel using the following command in your terminal:

ffmpeg -i sample_audio.wav -ac 1 -ar 16000 sample_audio_infer_ready.wav

Advanced Usage

Inference using CTC decoder

model.cur_decoder = "ctc"
ctc_text = model.transcribe(['sample_audio_infer_ready.wav'], batch_size=1, logprobs=False, language_id='ne')[0]
print(ctc_text)

Inference using RNNT decoder

model.cur_decoder = "rnnt"
rnnt_text = model.transcribe(['sample_audio_infer_ready.wav'], batch_size=1, language_id='ne')[0]
print(rnnt_text)

📄 License

This model is released under the MIT license.