ASR Language Modeling

[Pri4tam]

Describe the bug Aim: To Imporve the ASR model Accuracy with the help of LM. ASR Model Used: Nvidia Nemo: Conformer-CTC-BPE.nemo As per my understanding steps to implement ASR Language Modeling

1. Create KenLM Library.
2. Install Beam Search Decoder
3. Train N-Gram LM using ASR Model and KenLM Library. #output of step 3 is we have train N-Gram Model which can be used in beam search decoder on top of ASR model. (Decoder used PyCTCdecoder)
4. Evaluate by beam search decoder and N-Gram LM ##To tune parameter of beam search decoder. My question is after performing this steps how can I used my trained N-Gram Model with beam search decoder on top of ASR model.

Steps/Code to reproduce bug

Please list minimal steps or code snippet for us to be able to reproduce the bug.

A helpful guide on on how to craft a minimal bug report http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports.

Expected behavior

A clear and concise description of what you expected to happen.

Environment overview (please complete the following information)

• Environment location: [Bare-metal, Docker, Cloud(specify cloud provider - AWS, Azure, GCP, Collab)]
• Method of NeMo install: [pip install or from source]. Please specify exact commands you used to install.
• If method of install is [Docker], provide docker pull & docker run commands used

Environment details

If NVIDIA docker image is used you don't need to specify these. Otherwise, please provide:

• OS version
• PyTorch version
• Python version

Additional context

Add any other context about the problem here. Example: GPU model

[titu1994]

These docs may help - https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/stable/asr/asr_language_modeling.html

[Pri4tam]

I have referred this site only but not able to figure out the way for my mention problem. Can you please be specific on it.

[vsl9]

The evaluation script https://github.com/NVIDIA/NeMo/blob/stable/scripts/asr_language_modeling/ngram_lm/eval_beamsearch_ngram.py includes inference code and can save predictions if preds_output_folder argument is given.

[github-actions[bot]]

This issue is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.

[github-actions[bot]]

This issue was closed because it has been inactive for 7 days since being marked as stale.