ASR inference time too long

I am trying to test the ASR model. I uploaded sample file (10 second long .wav file) and clicked compute. The page just says compute loading and does not provide output. Am I mistaken with the type of audio file to be uploaded or should I check for whether the audio should be 8 or 16KHz file?

1 Like

Regarding to ASR config, I believe we need to submit 16KHZ file
ref: ESPnet2 pretrained model, kamo-naoyuki/mini_an4_asr_train_raw_bpe_valid.acc.best, fs=16k, lang=en

1 Like