Open spaco67 opened 2 months ago
Hello Bourke,
I'm very happy to see your concern on this matter, while I'm still on searching about this issue, and I'm really blindly looking for this work to be done, I'm studying software engineering, and I wanted to see this been done.
I don't know where are you, and what you're doing, but I'm looking forward discuss this matter with you so that we can join hand together to work on this matter.
I'm looking forward to your reply and you can also contact me on this email: asakrg@outlook.com
Thank you.
In addition to that, I also make a lot of research and I found out someone has been started fine-turning the Whisper to Hausa Language and I forked the project, and let's work together on it.
Hi Bourke,
I'm working on a project that involves using OpenAI's Whisper model, and I'm particularly interested in fine-tuning it with custom datasets for Hausa, a low-resource language. I believe this would greatly benefit communities working with underrepresented languages.
Objectives:
Fine-tune Whisper for the Hausa language: Steps or guidance on how to adapt the model using our own Hausa datasets. Enable both Speech-to-Text (STT) and Text-to-Speech (TTS) functionalities: Making the model versatile for various applications. Best practices for dataset preparation: Tips on formatting and annotating custom Hausa datasets to achieve optimal results.
Questions:
Additional Context:
Platform: Hugging Face Transformers Model in Use: OpenAI Whisper Language Focus: Hausa
Any guidance, suggestions, or references would be highly appreciated!
Thank you for your support.