Unable to transcribe on Linux

[carlganz]

predict(whisper_model, newdata="inst/demo_recording.wav", language = "en")

demo_recording.zip

Works perfectly on Windows, but on latest Ubuntu I get an inscrutable error: Error: tinyformat: Not enough conversion specifiers in format string

Any guidance would be greatly appreciated

[jwijffels]

is that a 16-bit .wav file?

[carlganz]

demo_recording.zip

I sent wrong file apologies

[jwijffels]

The wav file you are sending is not a 16-bit wav file

> library(av)
> av_media_info("C:/Users/jwijf/OneDrive/Bureaublad/demo_recording.wav")
$duration
[1] 24.64

$video
NULL

$audio
  channels sample_rate     codec frames bitrate     layout
1        1        8000 pcm_s16le     NA  128000 1 channels

[carlganz]

Interesting. Changing sampling rate did the trick. Surprised it worked at all on Windows.

I would strongly encourage adding some sort of input validation with clearer error messages.

Thank you for the quick response and helpful package.