Picovoice / eagle

On-device speaker recognition engine powered by deep learning
Apache License 2.0
23 stars 2 forks source link

Eagle Issue: NotImplementedError: Unsupported CPU: `0xd42`. #67

Closed bryanhughes closed 3 months ago

bryanhughes commented 3 months ago

Have you checked the docs and existing issues?

SDK

Python

Eagle package version

1.0.0

Framework version

Python 3.10.2

Platform

NVIDIA Jetson

OS/Browser version

Ubuntu 22.04

Describe the bug

I am on a Jetson Orin AGX developer kit with Jetpack 6.0GA. I am trying to run the demo but get the following:

$ eagle_demo_mic enroll --access_key ${ACCESS_KEY} --output_profile_path enrollments/
Traceback (most recent call last):
  File "/mnt/nvme/git/mimzy/mimzy-env/bin/eagle_demo_mic", line 8, in <module>
    sys.exit(main())
  File "/mnt/nvme/git/mimzy/mimzy-env/lib/python3.10/site-packages/pveagledemo/eagle_demo_mic.py", line 136, in main
    eagle_profiler = pveagle.create_profiler(
  File "/mnt/nvme/git/mimzy/mimzy-env/lib/python3.10/site-packages/pveagle/_factory.py", line 71, in create_profiler
    library_path = default_library_path()
  File "/mnt/nvme/git/mimzy/mimzy-env/lib/python3.10/site-packages/pveagle/_util.py", line 67, in default_library_path
    linux_machine = _linux_machine()
  File "/mnt/nvme/git/mimzy/mimzy-env/lib/python3.10/site-packages/pveagle/_util.py", line 47, in _linux_machine
    raise NotImplementedError("Unsupported CPU: `%s`." % cpu_part)
NotImplementedError: Unsupported CPU: `0xd42`.

It seems like this should support the Jetson as it is one of the Platform dropdowns, but perhaps it is out of date?

Steps To Reproduce

  1. $ eagle_demo_mic enroll --access_key ${ACCESS_KEY} --output_profile_path enrollments/

Expected Behavior

The demo would work

ErisMik commented 3 months ago

Unfortunately, this is expected behaviour. As per our docs in the README and online, the only NVIDIA Jetson platform we officially support is the NVIDIA Jetson Nano. That's what the dropdown in the bug report refers to. Apologies for the confusion.

There is a workaround where you could potentially use libraries designed for Raspberry Pi or the Jetson Nano, but there's no guarantee they will function either optimally or at all on your platform. Please understand that we cannot officially support the NVIDIA Jetson Orin AGX for the community at this time.

However, if this is for a commercial product and you're prepared for a more formal engagement, you can contact our sales team to inquire about this support through an enterprise package.