pythaiasr
pythaiasr copied to clipboard
Python Thai Automatic Speech Recognition
I test with 1:00 min wav file but it run out of GPU memory. > torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 138.00 MiB (GPU 0; 5.93 GiB total...
Add inputs as waveform arrays and inputs as batch to faster processing while use GPU.
Hi :) Is there a way to pipe audio from a soundcard directly into the `asr`?
Add Thai Whisper - https://huggingface.co/biodatlab/whisper-small-th-combined - https://huggingface.co/biodatlab/whisper-th-medium-combined - https://huggingface.co/biodatlab/whisper-th-large-combined