spleeter: ffprobe Error
Description
I downloaded homebrew and ffmpeg to run the code. Works fine with the audio_example given, but when I try to use another song from my library I get an ffprobe error. Also I’m putting this song in the same directory as the audio_example.mp3.
Step to reproduce
- Installed using
git clone https://github.com/Deezer/spleeter conda env create -f spleeter/conda/spleeter-cpu.yaml conda activate spleeter-cpu
- Run as
spleeter separate -i spleeter/Home.mp3 -p spleeter:2stems -o output
- Got
INFO:spleeter:Loading audio b'spleeter/Home.mp3' from 0.0 to 600.0 WARNING:spleeter:ffprobe error (see stderr output for detail)
error
Output
The code works fine with the original audio given. This would be the output.
INFO:spleeter:Audio data loaded successfully
INFO:spleeter:File output/audio_example/vocals.wav written
INFO:spleeter:File output/audio_example/accompaniment.wav written
Environment
OS | MacOS |
Installation type | Conda |
RAM available | 4 GB |
Hardware spec | CPU |
Additional context
About this issue
- Original URL
- State: closed
- Created 5 years ago
- Reactions: 1
- Comments: 28
ok guys… so I don’t know if this will help anyone or not - but I was researching, installing all sorts of shit (i’m a newbie here) over and over and kept replicating the same problem that we are all having.
I installed via pip, homebrew, and everyway I could follow or find but with no results.
I did the same with ffmpeg using above methods (homebrew etc) still same result.
Then I noticed another folder called spleeter INSIDE the spleeter folder that had been created by the installs. When I shifted my file into that folder ie. ‘/User/###/spleeter/spleeter’ - it worked without a hitch. no more this ffprobe or 0-6.00 errors and the rest. It just worked. Now, maybe this is the way it was meant to work all along, but with respect to the creators/devs, the manual didn’t mention this? (put the file in spleeter/spleeter) or perhaps it did… but with the complicated installs and jumping around webpages… i could have missed it. The odd thing, is that the audio example is still in the top folder… so IDK -wtf is going on. lol. The guide is not easy for newbie people who’ve never used command lines, things like pip or homebrew and terminal concepts. But I did get it working and my troubles were rewarded, but still don’t know if it was luck or a solution…?. hopefully, it may help or steer others and get it working for them. I included a screenshot in hopes it may help… but I’ve done about 10 rips at 2 track and 5 track and its bloody worth the effort. Thank you so much to the developers for this! but… can we make some sort of graphical interface or an easier streamlined process would be amazing!!
Much respect and honour to the devs/users, and i’m crossing fingers this just may work for someone. Cheers.
Hi
Got the same issue with docker as well
I did an initial install of CPU version and successfully stemmed a few tracks/. Installed the GPU and abandoned it due to being on a VM. Went back to CPU and received the ffprobe error. Came here read up and in stalled via (git clone https://git.ffmpeg.org/ffmpeg.git ffmpeg )
Not only did it FIX the ffprobe error but I feel like the quality of the sounds of the stems vastly improved.
Ok thanks let me also see further into this
Sent from my iPhone
Ok this is indeed the expected behavior. the separation command goes:
spleeter separate -i <path_to_audio_file> -o <path_to_audio_output_dir>
where thepath_to_audio_file
parameter must be a valid path and depends on where you are executing the command.if your audio file
home.mp3
is in the current directory (the directory in which you are executing the command) then the right call is justspleeter separate -i home.mp3
. The additionnalspleeter
here in the example provided is to indicate that the audio file is located there.For Docker, you can solve this problem by reading the section that starts with ‘For using your own audio file you will need to create contain…’ on the Usage page of the wiki