search

mustafaaljadery / lightning-whisper-mlx

An extremely fast implementation of whisper optimized for Apple Silicon using MLX.
https://mustafaaljadery.github.io/lightning-whisper-mlx/
469 stars 21 forks source link

what(): [Matmul::eval_cpu] Currently only supports float32. #3

Closed gyokuro33 closed 3 months ago

gyokuro33 commented 3 months ago

The following error occurred when executing the code. What could be the possible reasons? Reading the error statement, it appears that somewhere in the library a mistake has been made in the type of calculation. thank you.

code: `from lightning_whisper_mlx import LightningWhisperMLX import time

whisper = LightningWhisperMLX(model="large-v3", batch_size=12, quant=None) time_sta = time.time() text = whisper.transcribe(audio_path="asano.wav")['text'] time_end = time.time() tim = time_end - time_sta print(text) print(tim)`

error: terminate called after throwing an instance of 'std::runtime_error' what(): [Matmul::eval_cpu] Currently only supports float32. Aborted (core dumped)

gyokuro33 commented 3 months ago

I'm sorry. I was trying to use it on ubuntu without realizing it was for Mac.