Instructions to use Audio8/ARK-ASR-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Audio8/ARK-ASR-3B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="Audio8/ARK-ASR-3B", trust_remote_code=True)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("Audio8/ARK-ASR-3B", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Please support for standard arabic speech to text (with harakat tashkil Diacritics if better) 🙏🏻
Amazing model and it's so good on benchmarks, i wish you can support arabic speech to text as it so poor in open source community and have high demand
Thank you for your interest and kind words! One of the challenges is the limited availability of large-scale, high-quality open Arabic speech datasets, especially with reliable diacritized transcripts. Nevertheless, we’re activelyly collecting more diverse data and plan to gradually expand support to additional languages in future releases. Stay tuned!
Thank you for your response, this is a 3000 hours very rich arabic speech dataset with dialect "oddadmix/dialectal-arabic-lahgtna-v2"
And this is specific moroccan rich speech dataset "atlasia/MoulSot-Full" 1,500 hours