o//whisper.cpp/main: No such file or directory

[sujoykb]

What happened?

I cloned the latest llamafile repo and tried to run whisperfile following this readme:

make -j8 o//whisper.cpp
o//whisper.cpp/main -m whisper-tiny.en-q5_1.bin -f whisper.cpp/jfk.wav --no-prints

Make did not show any error but the second command failed with this message: o//whisper.cpp/main: No such file or directory

I tried this on Google Colab (w/ and w/o GPU) as well as on my local machine (no GPU). Nowhere I was able to find any executable main inside o//whisper.cpp.

The following files are there inside o//whisper.cpp (showing you the output of Google colab below):

Version

llamafile 0.8.12 (latest)

What operating system are you seeing the problem on?

Linux

Relevant log output

No response

[artyom]

I ran into the same issue, and for me the following change did the trick:

diff --git whisper.cpp/BUILD.mk whisper.cpp/BUILD.mk
index 7aa96f1..8daab0a 100644
--- whisper.cpp/BUILD.mk
+++ whisper.cpp/BUILD.mk
@@ -43,4 +43,5 @@ $(WHISPER_CPP_OBJS): whisper.cpp/BUILD.mk

 .PHONY: o/$(MODE)/whisper.cpp
 o/$(MODE)/whisper.cpp:                 \
+       o/$(MODE)/whisper.cpp/main \
        o/$(MODE)/whisper.cpp/server

[sujoykb]

>  .PHONY: o/$(MODE)/whisper.cpp
>  o/$(MODE)/whisper.cpp:                   \
> +     o/$(MODE)/whisper.cpp/main \
>       o/$(MODE)/whisper.cpp/server

Correct, this works! Thank you so much @artyom BTW, have you been able to run the whisperfile with cuda? Adding --gpu auto did not work for me (tried on colab).

[jart]

I've committed that change to the repository. Thanks for the report. For the record, you could also say make -j o//whisper.cpp/main. I've also updated things so that make -j && sudo make install will create an executable on your PATH called whisperfile. Enjoy!