Datasets:
Labeling training data
@freococo thanks so much for collecting and sharing this audio dataset, it’s a big contribution to Rohingya language resources.
I'm interested in training ASR models for Rohingya and want to avoid duplicating effort. Have you (or anyone you know) already done any labeling/transcription work for the audio in this repo, or started a plan to do so (even partially)?
If there are existing transcripts, alignment files, annotation guidelines, or a preferred labeling format, we’d love to coordinate and build on what’s already underway. Thanks again!
Thanks a lot for your kind words and interest 🙏
To be fully transparent, I personally don’t speak Rohingya, so I haven’t done any transcription or labeling work on this audio yet, and there’s no existing transcript, alignment, or annotation plan in place from my side.
That said, I’d be very happy if you (or others working on ASR) want to download the dataset and add transcripts or annotations. If you start any transcription or develop annotation guidelines or a preferred format, I’d love to coordinate and link or include that work with the repo so others can build on it rather than duplicate effort.
Thanks again, and feel free to open an issue or discussion if you want to align on formats or next steps.