Arabic language in Qwen3-TTS
I am working on the Qwen3-TTS model and I have improved the Arabic language in it. Is it possible for us to work together to train it on your data?
hi @sherif1313 , let me know more details and what are you looking to do and lets figure it out.
Are we you trying to finetune QwenTTS for specific dialact or what do you have in mind?
Also do you have the GPU resources?
hi ahmed
I previously translated the GLM-TTS model into Arabic, and before that, I used this Chatterbox model you're working on because it's not open source and only works through LoRa. I also translated the index-tts model into Arabic, but I found them to be very large and requiring a lot of data. Since the internet in Egypt is very weak, I couldn't train it on such a large dataset. So, I translated the Qwen3-TTS model, but my data is very limited, and it also needs a large amount of data to translate Arabic well. Therefore, I suggest we translate one of these models using your own data, creating a separate, open-source model, unlike the closed Chatterbox model.
Chatterbox is not closed source. it's open source and under MIT license.
Yes, I've now confirmed that it's open source, but I trained it on data and couldn't integrate the model, which is why I thought it was closed-source.
https://github.com/stlohrey/chatterbox-finetuning/
Could you please tell me how to integrate the model with Lora?