Skip to content

Update Parakeet ASR Fine-tune Tutorials #223

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
kingformatty wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Update Parakeet ASR Fine-tune Tutorials #223

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
56 changes: 41 additions & 15 deletions asr-finetune-parakeet-nemo.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,17 +65,42 @@
" a. Click **Runtime** > **Restart Runtime** for any upgraded packages to take effect.\n",
"\"\"\"\n",
"\n",
"# Before running this notebook\n",
" # pull docker container: docker pull nvcr.io/nvidia/cuda:12.4.1-cudnn-devel-ubuntu22.04 (recommended)\n",
" # launch container and run below\n",
" # `apt-get update`\n",
" # install python3.10.12 and pip\n",
" # `apt-get install -y --no-install-recommends python3.10 python3.10-venv python3.10-distutils python3.10-dev`\n",
" # `apt-get install -y --no-install-recommends python3-pip`\n",
" # `update-alternatives --install /usr/bin/python python /usr/bin/python3.10 1`\n",
" # `update-alternatives --set python /usr/bin/python3.10`\n",
" # `python --version`\n",
"\n",
"# run this notebook (you might need to run this cell two times where the site-packages need to be synced in first run)\n",
"\n",
"# Install Dependencies\n",
"!pip install wget\n",
"!apt-get install sox libsndfile1 ffmpeg libsox-fmt-mp3 jq\n",
"!pip install text-unidecode\n",
"!pip install matplotlib>=3.3.2\n",
"!pip install Cython\n",
"!pip3 install --no-cache-dir huggingface-hub==0.23.2\n",
"!pip3 install wget\n",
"!apt-get install -y wget sox libsndfile1 ffmpeg libsox-fmt-mp3 jq git # if failed, run `apt-get update` first\n",
"!pip3 install text-unidecode\n",
"!pip3 install matplotlib\n",
"!pip3 install Cython\n",
"!pip3 install torch==2.3.0\n",
"!pip3 install librosa\n",
"\n",
"\n",
"## Install NeMo\n",
"BRANCH = 'v1.23.0'\n",
"!python -m pip install git+https://github.com/NVIDIA/NeMo.git@$BRANCH#egg=nemo_toolkit[all]\n",
"# Install NeMo\n",
"BRANCH = 'v2.0.0'\n",
"!python3 -m pip install git+https://github.com/NVIDIA/NeMo.git@$BRANCH#egg=nemo_toolkit[asr]\n",
"\n",
"# Modify some package for dependency consideration\n",
"!pip3 install numpy==1.24.1\n",
"!pip3 install --no-cache-dir huggingface-hub==0.24.5\n",
"!pip3 install transformers==4.40.2\n",
"!pip3 install numba==0.59.0\n",
"!pip3 install pytorch-lightning==2.3.3\n",
"!pip3 install lightning-utilities==0.10.1\n",
"!pip3 install omegaconf==2.3.0\n",
"!pip3 install jiwer==3.0.4\n",
"\n",
"\"\"\"\n",
"Remember to restart the runtime for the kernel to pick up any upgraded packages (e.g. matplotlib)!\n",
Expand Down Expand Up @@ -300,8 +325,9 @@
"source": [
"# To fully train the model, you'll need to increase trainer.max_epochs from 1.\n",
"# Empirical evidence suggests that around 200 epochs should suffice.\n",
"NEMO_DIR = 'FIX_ME/path/to/NeMo'\n",
"! git clone -b $BRANCH https://github.com/NVIDIA/NeMo $NEMO_DIR\n",
"NEMO_DIR = '/workspace/NeMo'\n",
"BRANCH='v2.0.0'\n",
"!git clone -b $BRANCH https://github.com/NVIDIA/NeMo $NEMO_DIR\n",
"!python $NEMO_DIR/examples/asr/speech_to_text_finetune.py \\\n",
" --config-path=\"../asr/conf/fastconformer/hybrid_transducer_ctc/\" --config-name=fastconformer_hybrid_transducer_ctc_bpe \\\n",
" +init_from_pretrained_model=stt_en_fastconformer_hybrid_large_pc \\\n",
Expand All @@ -322,7 +348,7 @@
},
{
"cell_type": "code",
"execution_count": null,
"execution_count": 9,
"metadata": {
"id": "_w2KBGHaSf4S"
},
Expand Down Expand Up @@ -417,7 +443,7 @@
"source": [
"!pip install nvidia-pyindex\n",
"!ngc registry resource download-version \"nvidia/riva/riva_quickstart:\"$__riva_version__\n",
"!pip install nemo2riva\n",
"!pip install nemo2riva==2.19.0\n",
"!pip install protobuf==3.20.0"
]
},
Expand All @@ -443,7 +469,7 @@
"outputs": [],
"source": [
"riva_file_path = ctc_model_path[:-5]+\".riva\"\n",
"!nemo2riva --key=nemotoriva --onnx-opset 18 --out $riva_file_path $ctc_model_path"
"!nemo2riva --key=nemotoriva --max-dim 1000 --onnx-opset 18 --out $riva_file_path $ctc_model_path"
]
},
{
Expand Down Expand Up @@ -482,7 +508,7 @@
},
"gpuClass": "standard",
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
Expand Down