[Bug]: audacity.fetch-models: command not found

[tomasris]

What happened?

on debian:
sudo snap install audacity --channel latest/candidate
sudo snap install intel-npu-driver
sudo usermod -a -G render $USER
sudo audacity.fetch-models
sudo: audacity.fetch-models: command not found
sudo: fetch-models: command not found

What should have happened?

will downloaded models for OpenVino

Output of snap info $snap_name

snap info audacity
name:      audacity
summary:   Audacity is the world's most popular audio editing and recording app
publisher: Snapcrafters✪
store-url: https://snapcraft.io/audacity
contact:   https://github.com/snapcrafters/audacity/issues
license:   MIT
description: |
  Audacity® is a free, easy-to-use, multi-track audio editor and recorder for Windows,
  Mac OS X, GNU/Linux and other operating systems. The interface is translated into
  many languages.
  
  You can use Audacity to:
  * Record live audio
  * Convert tapes and records into digital recordings or CDs
  * Edit WAV, AIFF, FLAC, MP2, MP3 or Ogg Vorbis sound files
  * Cut, copy, splice or mix sounds together
  * Change the speed or pitch of a recording
  * Apply a wide range of other effects to audio recordings
  
  This snap also includes OpenVINO™ AI plugins for music separation, noise suppression,
  music generation and continuation, transcription, and super resolution. These plugins
  can only be run on Intel hardware (CPU, GPU, and NPU) and a user must be in the render
  Unix group in order to run on Intel accelerators (GPU and NPU). To ease the installation
  of the AI models, the snap also includes a model management command that can be run using:
  
  sudo audacity.fetch-models
  
  For Intel NPU support, make sure to also run the following prior to running Audacity:
  
  sudo snap install intel-npu-driver
  sudo snap connect audacity:npu-libs intel-npu-driver:npu-libs
commands:
  - audacity
  - audacity.fetch-models
snap-id:      KTe2wdAu5JKdRDUgYBuXXGjDXyzobvFI
tracking:     latest/candidate
refresh-date: today at 22:43 BST
channels:
  latest/stable:    3.7.1 2025-06-16 (1208) 641MB -
  latest/candidate: 3.7.1 2025-06-16 (1208) 641MB -
  latest/beta:      ↑                             
  latest/edge:      3.7.1 2025-06-17 (1210) 641MB -
installed:          3.7.1            (1208) 641MB -

Output of snap connections $snap_name

snap connections audacity
Interface               Plug                      Slot                            Notes
alsa                    audacity:alsa             :alsa                           manual
audio-playback          audacity:audio-playback   :audio-playback                 -
audio-record            audacity:audio-record     -                               -
content[ffmpeg-2404]    audacity:ffmpeg-2404      ffmpeg-2404:ffmpeg-2404         -
content[gnome-46-2404]  audacity:gnome-46-2404    gnome-46-2404:gnome-46-2404     -
content[gpu-2404]       audacity:gpu-2404         mesa-2404:gpu-2404              -
content[gtk-3-themes]   audacity:gtk-3-themes     gtk-common-themes:gtk-3-themes  -
content[icon-themes]    audacity:icon-themes      gtk-common-themes:icon-themes   -
content[npu-libs-2404]  audacity:npu-libs         intel-npu-driver:npu-libs       -
content[sound-themes]   audacity:sound-themes     gtk-common-themes:sound-themes  -
custom-device           audacity:intel-npu        intel-npu-driver:intel-npu      -
desktop                 audacity:desktop          :desktop                        -
desktop-legacy          audacity:desktop-legacy   :desktop-legacy                 -
gsettings               audacity:gsettings        :gsettings                      -
home                    audacity:home             :home                           -
jack1                   audacity:jack1            -                               -
network                 audacity:network          :network                        -
network-bind            audacity:network-bind     :network-bind                   -
opengl                  audacity:opengl           :opengl                         -
pulseaudio              audacity:pulseaudio       :pulseaudio                     -
removable-media         audacity:removable-media  -                               -
unity7                  audacity:unity7           :unity7                         -
wayland                 audacity:wayland          :wayland                        -
x11                     audacity:x11              :x11                            -

Output of snap version

snap    2.70

Relevant log output

Teminal output of app

From root manual created directory common and tmp, and start command from root, started to work, but I don't think it's a right way.

[frenchwr]

Hi @tomasris , to be honest I have never tried running the snap on debian. It appears that the command is not on your command search path (PATH environment variable) when you run with sudo. The command should be installed in the same directory as the audacity command. Something like /snap/bin.

Once you find the path to the command then try passing that path with sudo. For example, here is the result on my system (Ubuntu 24.04):

$ which  audacity.fetch-models 
/snap/bin/audacity.fetch-models

$ sudo /snap/bin/audacity.fetch-models

-------------------------------------------------------------------------------------------------
| Audacity OpenVINO™ plugins support several AI models available for download from Hugging Face |
-------------------------------------------------------------------------------------------------

1. Music Generation
2. Whisper Transcription
3. Music Separation
4. Noise Suppression
5. Super Resolution
6. Exit

Some models are several GB in size, so be mindful of your available disk space and network speed.

To download and install a model, please select an option from the menu above and hit enter:

If the first command does not return anything, then you could try search for it with find. Something like:

find /snap -name audacity.fetch-models

Let me know if that helps.

[tomasris]

Yes, it worked, or partially (sometimes you have to create the common/tmp directory manually, option -v helping). Thank you. However, the models were downloaded, but Audacity does not see it. Perhaps need to specify where to look for this directory /var/snap/audacity/common/models/openvino-models ?
snap? There are no alternative options. How can I install OpenVINO + audacity on Debian in less than an hour? So convenient, simple install/uninstall, and safe

[frenchwr]

Hi @tomasris , I suspect that the models did not completely download or install correctly. Let's focus on Music Generation first.

Does the fetch-models menu report that the models are installed?

1. Music Generation (installed)

If yes, do you see the following files on your system?

$ tree /var/snap/audacity/common/models/openvino-models/musicgen/
/var/snap/audacity/common/models/openvino-models/musicgen/
├── attention_mask_from_prepare_4d_causal_10s.raw
├── attention_mask_from_prepare_4d_causal_5s.raw
├── encodec_20s.xml
├── encodec_5s.xml
├── encodec_combined_weights.bin
├── encodec_encoder_10s.xml
├── encodec_encoder_5s.xml
├── encodec_encoder_combined_weights.bin
├── mono
│   ├── embed_tokens.bin
│   ├── embed_tokens.xml
│   ├── enc_to_dec_proj.bin
│   ├── enc_to_dec_proj.xml
│   ├── initial_cross_attn_kv_producer.bin
│   ├── initial_cross_attn_kv_producer.xml
│   ├── lm_heads.bin
│   ├── lm_heads.xml
│   ├── musicgen_decoder_combined_weights.bin
│   ├── musicgen_decoder_combined_weights_int8.bin
│   ├── musicgen_decoder_static0_10s.xml
│   ├── musicgen_decoder_static0_5s.xml
│   ├── musicgen_decoder_static_batch1_int8.xml
│   ├── musicgen_decoder_static_batch1.xml
│   ├── musicgen_decoder_static_int8.xml
│   ├── musicgen_decoder_static.xml
│   └── sinusoidal_positional_embedding_weights_2048_1024.raw
├── musicgen-small-tokenizer.bin
├── musicgen-small-tokenizer.xml
├── stereo
│   ├── embed_tokens.bin
│   ├── embed_tokens.xml
│   ├── enc_to_dec_proj.bin
│   ├── enc_to_dec_proj.xml
│   ├── initial_cross_attn_kv_producer.bin
│   ├── initial_cross_attn_kv_producer.xml
│   ├── lm_heads.bin
│   ├── lm_heads.xml
│   ├── musicgen_decoder_combined_weights.bin
│   ├── musicgen_decoder_combined_weights_int8.bin
│   ├── musicgen_decoder_static0_10s.xml
│   ├── musicgen_decoder_static0_5s.xml
│   ├── musicgen_decoder_static_batch1_int8.xml
│   ├── musicgen_decoder_static_batch1.xml
│   ├── musicgen_decoder_static_int8.xml
│   ├── musicgen_decoder_static.xml
│   └── sinusoidal_positional_embedding_weights_2048_1024.raw
├── t5.bin
└── t5.xml

3 directories, 46 files

And how much space on disk do the models use? Here is the result on my system:

$ du -sh /var/snap/audacity/common/models/openvino-models/musicgen/
2.8G	/var/snap/audacity/common/models/openvino-models/musicgen/

[tomasris]

Music Generation models (~2.8GB) successfully installed to /var/snap/audacity/common/models/openvino-models/musicgen
Whisper Transcription models (~1.5GB) successfully installed to /var/snap/audacity/common/models/openvino-models
Music Separation models (~99MB) successfully installed to /var/snap/audacity/common/models/openvino-models
Downloading Noise Suppression models from Hugging Face. Please be patient!
Failed to Fopen file /etc/wgetrc
wget: Cannot read /etc/wgetrc (Permission denied).
Super Resolution models (~2GB) successfully installed to /var/snap/audacity/common/models/openvino-models/audiosr
2cond attempt
Noise Suppression models (~27MB) successfully installed to /var/snap/audacity/common/models/openvino-models

sudo tree /var/snap/audacity/common/models/openvino-models/musicgen/
/var/snap/audacity/common/models/openvino-models/musicgen/
├── attention_mask_from_prepare_4d_causal_10s.raw
├── attention_mask_from_prepare_4d_causal_5s.raw
├── encodec_20s.xml
├── encodec_5s.xml
├── encodec_combined_weights.bin
├── encodec_encoder_10s.xml
├── encodec_encoder_5s.xml
├── encodec_encoder_combined_weights.bin
├── mono
│   ├── embed_tokens.bin
│   ├── embed_tokens.xml
│   ├── enc_to_dec_proj.bin
│   ├── enc_to_dec_proj.xml
│   ├── initial_cross_attn_kv_producer.bin
│   ├── initial_cross_attn_kv_producer.xml
│   ├── lm_heads.bin
│   ├── lm_heads.xml
│   ├── musicgen_decoder_combined_weights.bin
│   ├── musicgen_decoder_combined_weights_int8.bin
│   ├── musicgen_decoder_static0_10s.xml
│   ├── musicgen_decoder_static0_5s.xml
│   ├── musicgen_decoder_static_batch1_int8.xml
│   ├── musicgen_decoder_static_batch1.xml
│   ├── musicgen_decoder_static_int8.xml
│   ├── musicgen_decoder_static.xml
│   └── sinusoidal_positional_embedding_weights_2048_1024.raw
├── musicgen-small-tokenizer.bin
├── musicgen-small-tokenizer.xml
├── stereo
│   ├── embed_tokens.bin
│   ├── embed_tokens.xml
│   ├── enc_to_dec_proj.bin
│   ├── enc_to_dec_proj.xml
│   ├── initial_cross_attn_kv_producer.bin
│   ├── initial_cross_attn_kv_producer.xml
│   ├── lm_heads.bin
│   ├── lm_heads.xml
│   ├── musicgen_decoder_combined_weights.bin
│   ├── musicgen_decoder_combined_weights_int8.bin
│   ├── musicgen_decoder_static0_10s.xml
│   ├── musicgen_decoder_static0_5s.xml
│   ├── musicgen_decoder_static_batch1_int8.xml
│   ├── musicgen_decoder_static_batch1.xml
│   ├── musicgen_decoder_static_int8.xml
│   ├── musicgen_decoder_static.xml
│   └── sinusoidal_positional_embedding_weights_2048_1024.raw
├── t5.bin
└── t5.xml

3 directories, 46 files

du -sh /var/snap/audacity/common/models/openvino-models/musicgen/
2.8G /var/snap/audacity/common/models/openvino-models/musicgen/

[frenchwr]

Failed to Fopen file /etc/wgetrc
wget: Cannot read /etc/wgetrc (Permission denied).

This looks like an actual issue, I cannot reproduce it so it may be something specific to debian, or perhaps in my local environment (though I don't believe this is the case). We'll have to come back to this. It only affects noise separation.

du -sh /var/snap/audacity/common/models/openvino-models/musicgen/
2.8G /var/snap/audacity/common/models/openvino-models/musicgen/

This looks correct, so the models appear to be installed. When you open the Music Generation window in Audacity, what do you see under Model Selection? Is it empty? Here is what I see on my system:

Could you share the Audacity logs (Help -> Diagnostics -> Show Logs) after opening and closing the Music Generation window?

Could you also run (if you are not already) Audacity from the command line (with the audacity command) and share the logs you see there?

Also, is your user in the render group on your machine? You can run groups to check.

[tomasris]

OpenVINO Music Generation V2 has no models installed.

[tomasris]

groups

• cdrom floppy sudo audio dip video plugdev users netdev scanner bluetooth lpadmin render

$ audacity
libpxbackend-1.0.so: cannot open shared object file: No such file or directory
Failed to load module: /home/*/snap/audacity/common/.cache/gio-modules/libgiolibproxy.so
g_module_open() failed for /snap/audacity/1208/gnome-platform/usr/lib/x86_64-linux-gnu/gdk-pixbuf-2.0/2.10.0/loaders/libpixbufloader_svg.so: /snap/audacity/1208/gnome-platform/usr/lib/x86_64-linux-gnu/gdk-pixbuf-2.0/2.10.0/loaders/libpixbufloader_svg.so: undefined symbol: rsvg_handle_get_pixbuf_and_error
Testing for explicit PulseAudio choice...
Testing for ALSA permissions...
...and using ALSA.

[tomasris]

Finally. I deleted all openvino installations via snap, and that helped.

I manually copied the downloaded models to /var/snap/audacity/common/models/openvino-models, and after launching Audacity, openvino found the models. Most likely, somewhere in the paths there was a setting to search for models in a different location, and after deleting it, everything worked fine. I don't have time to regenerate the cause of the malfunction. Thank you.
I also excluded working with ALSA, but that's another story.