Join the community of Machine Learners and AI enthusiasts.
Sign Up
Basically, the HF Spaces are for demonstration purposes. There is no problem using them for other than demonstrations, though.
The free CPU Spaces are slow because they don't have GPUs, and the Zero GPU Spaces have a quota, so in principle you have to finish processing within 120 seconds.😓
If someone were to release a space on a pay-as-you-go GPU plan, there would probably be no restrictions, but that would be quite rare.
For serious use, you should consider local environments or a paid service including other companies.
This answer not justify a fact to not exist any space or project able to transcribe a single 3 min audio file. Sorry but this service is very weak.
There are many whisper spaces here, which can probably easily transcribe 3 min files. Try this maybe: https://huggingface.co/spaces/KingNish/Realtime-whisper-large-v3-turbo
