-
Notifications
You must be signed in to change notification settings - Fork 215
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.
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
Adds radio data recipe #1400
base: master
Are you sure you want to change the base?
Adds radio data recipe #1400
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks! Please fix the CI and the torchaudio import.
lhotse/recipes/radio.py
Outdated
|
||
from lhotse.parallel import parallel_map | ||
from lhotse.audio import set_ffmpeg_torchaudio_info_enabled | ||
import torchaudio |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Make torchaudio imports local, it’s not a core dependency anymore.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think it should all be fixed now. I have an associated recipe in icefall for doing LID and geolocation that I am going to submit a pull request for. Part of that involves a custom dataset class, but it's fairly specific to this one task. I assume it would be better to just have that as part of the icefall recipe and not part of Lhotse?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah that makes sense.
from lhotse.supervision import SupervisionSegment, SupervisionSet | ||
from lhotse.utils import Pathlike | ||
|
||
set_ffmpeg_torchaudio_info_enabled(False) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
One last request - don’t set these things in the global scope because it’ll be executed when Lhotse is imported. Move it to local scope (even better i think this one works like a context manager)
This is the recipe for some radio data we collected and are able to distribute if someone emails us for it. We may possibly have a way of releasing it with some restrictions in the future, but for now it is just a prepare script. The paper it corresponds to is
https://aclanthology.org/2024.naacl-long.286.pdf
The main use is for geolocating speech, self-supervised model pretraining, or language id.