I purchased it to make a speech recognition server.
I originally intended to install what was built in centOS in JETSON.
I installed JETPACK.
It uses cuda and ldd for speech recognition.
However, glibc.i686 and libgomp.i686 are required but can not be installed.
In a centOS environment, it is too difficult to install even with eyes closed.
So I want to ask.
Does Jetson TX2 provide a speech recognition SDK?
Hi kwm6623, JetPack-L4T doesn’t directly feature a speech recognition SDK, however there are various open-source speech recognition products included in the Ubuntu apt package repo that you can install from the command line or Ubuntu Software Center. For example CMU Sphinx / pocketsphinx and Hey, Jetson!
These are the x86 runtimes for PC. Jetson TX1/TX2 are arm64 (aarch64) architecture. Do you have access to the source code? You will need to recompile it for Jetson / aarch64.
This means that the binary is compiled for Intel CPUs, not ARM CPUs.
You should switch over to an ARM CentOS install, if that exists.
Else, you should be able to download the sources and build it yourself on the Jetson.
I don’t understand what the dependency on “ldd” is though – “ldd” is a tool that shows what shared libraries are needed by other libraries. “ldd” exists (and glibc exists) on the stock Jetson install, in their ARM forms.
Are you trying to use some proprietary, binary-only, speech recognition library or tool? If so, you need one that’s compiled for AArch64 from your vendor.
If you can’t get that, have you considered Sphinx or Julius, which are open source projects?
Here’s an interesting speech recognition project for TX1/TX2 called Hey, Jetson!