User Tools

Site Tools


projects:jetson:orin:whisper:start

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
projects:jetson:orin:whisper:start [2023/03/13 20:55] – [1. Upgrade to Jetpack 5.1] rolf.beckerprojects:jetson:orin:whisper:start [2023/03/14 20:27] (current) – [Test OpenAI Whisper] rolf.becker
Line 7: Line 7:
  
 ===== About Whisper ===== ===== About Whisper =====
 +
 +  * https://openai.com/research/whisper
   * https://github.com/openai/whisper   * https://github.com/openai/whisper
   * https://towardsdatascience.com/transcribe-audio-files-with-openais-whisper-e973ae348aa7   * https://towardsdatascience.com/transcribe-audio-files-with-openais-whisper-e973ae348aa7
Line 26: Line 28:
 ===== Setup ===== ===== Setup =====
  
-==== 0. Install pip and virtualenv ====+==== 1. Upgrade to Jetpack 5.1 ==== 
 + 
 +This means basically a completely **new installation** / initial setup. 
 +\\ Follow [[https://developer.nvidia.com/embedded/learn/get-started-jetson-agx-orin-devkit|Getting Started with Jetson AGX Orin Developer Kit]] to flash the Orin new. 
 +\\ Use the [[https://developer.nvidia.com/drive/sdk-manager|NVIDIA SDK Manager]] to setup the Orin via USB from scratch.  
 + 
 + 
 + 
 +==== 2. Install pip and virtualenv for Python3 ====
  
 <code> <code>
Line 38: Line 48:
  
  
-==== 1. Upgrade to Jetpack 5.1 ==== 
- 
-This means basically a completely new installation / initial setup. 
-\\ Use the [[https://developer.nvidia.com/drive/sdk-manager|NVIDIA SDK Manager]] to setup the Orin via USB from scratch.  
  
-==== 2. Install PyTorch for Jetson ====+==== 3. Install PyTorch for Jetson ====
  
 :!: Follow {{https://docscontent.nvidia.com/dita/00000186-be6e-d774-adce-be7e84580000/deeplearning/frameworks/pdf/Install-PyTorch-Jetson-Platform.pdf|Install-PyTorch-Jetson-Platform.pdf}} :!: Follow {{https://docscontent.nvidia.com/dita/00000186-be6e-d774-adce-be7e84580000/deeplearning/frameworks/pdf/Install-PyTorch-Jetson-Platform.pdf|Install-PyTorch-Jetson-Platform.pdf}}
Line 52: Line 58:
  
 <code> <code>
-sudo apt-get -y update; +sudo apt-get -y update 
-sudo apt-get -y install autoconf bc build-essential g++-8 gcc-8 clang-8 lld-8 gettext- + 
-base gfortran-8 iputils-ping libbz2-dev libc++-dev libcgal-dev libffi-dev libfreetype6- +sudo apt-get -y install 
-dev libhdf5-dev libjpeg-dev liblzma-dev libncurses5-dev libncursesw5-dev libpng-dev +autoconf bc build-essential g++-8 gcc-8 clang-8 lld-8 gettext-base gfortran-8 
-libreadline-dev libssl-dev libsqlite3-dev libxml2-dev libxslt-dev locales moreutils +iputils-ping libbz2-dev libc++-dev libcgal-dev libffi-dev libfreetype6-dev libhdf5-dev 
-openssl python-openssl rsync scons python3-pip libopenblas-dev;+libjpeg-dev liblzma-dev libncurses5-dev libncursesw5-dev libpng-dev libreadline-dev libssl-dev 
 +libsqlite3-dev libxml2-dev libxslt-dev locales moreutils openssl python-openssl rsync scons 
 +python3-pip virtualenv libopenblas-dev
 </code> </code>
  
-**Check for the latest version:** \\ https://developer.download.nvidia.com/compute/redist/jp/v51/pytorch/+**Check for the latest version of PyTorch for Jetpack:** \\ https://developer.download.nvidia.com/compute/redist/jp/v51/pytorch/
  
 Save the result in an environment variable:\\ Save the result in an environment variable:\\
Line 68: Line 76:
  
  
-==== Install PyTorch in the environment nlp ====+ 
 +=== Install PyTorch in the environment nlp ===
  
 <code> <code>
 +# create and activate the environment nlp
 virtualenv -p python3 nlp virtualenv -p python3 nlp
 source nlp/bin/activate source nlp/bin/activate
-# pip install --upgrade pip; +
 # pip install aiohttp numpy=='1.19.4' scipy=='1.5.3'  # pip install aiohttp numpy=='1.19.4' scipy=='1.5.3' 
 pip install aiohttp numpy scipy  pip install aiohttp numpy scipy 
-export "LD_LIBRARY_PATH=/usr/lib/llvm-8/lib:$LD_LIBRARY_PATH"; python3 -m+export "LD_LIBRARY_PATH=/usr/lib/llvm-8/lib:$LD_LIBRARY_PATH"
 pip install --upgrade protobuf;  pip install --upgrade protobuf; 
 pip install --no-cache $TORCH_INSTALL pip install --no-cache $TORCH_INSTALL
 </code> </code>
  
-<code> +=== Install Jupyterlab and Pandas ===
-export TORCH_INSTALL=https://developer.download.nvidia.com/compute/redist/jp/v51/pytorch/torch-1.14.0a0+44dac51c.nv23.02-cp38-cp38-linux_aarch64.whl +
- +
-virtualenv -p python3 nlp +
- +
-source nlp/bin/activate +
- +
-#python3 -m pip install aiohttp numpy=='1.19.4' scipy=='1.5.3'  +
- +
-pip install aiohttp numpy scipy  +
- +
-export "LD_LIBRARY_PATH=/usr/lib/llvm-8/lib:$LD_LIBRARY_PATH"; python3 -m +
- +
-pip install --no-cache $TORCH_INSTALL+
  
 +<code>
 +pip install pandas matplotlib jupyterlab
 </code> </code>
  
  
 +=== Install Whisper by OpenAI ===
 <code> <code>
-pip install pandas matplotlib jupyterlab 
- 
 pip install -U openai-whisper pip install -U openai-whisper
- 
 </code> </code>
  
  
 +=== In case you want to delete the installation: ===
  
 <code> <code>
 deactivate deactivate
- 
 rm -Rf nlp rm -Rf nlp
 +</code>
 +
 +==== Test OpenAI Whisper ====
 +
 +<code>
 +source nlp/bin/activate
 +
 +pip install ipywidgets 
 +
 +pip install torchaudio
 +
  
 </code> </code>
projects/jetson/orin/whisper/start.1678737335.txt.gz · Last modified: 2023/03/13 20:55 by rolf.becker