maks/sonus
1
0
Fork 0
Code Issues Pull requests 1 Projects Releases Packages Wiki Activity Actions

Add karaoke mode with on-demand vocal separation via Demucs #2

Open
maks wants to merge 2 commits from feature/karaoke-mode into master
pull from: feature/karaoke-mode
merge into: maks:master
Conversation 0 Commits 2 Files changed 16 +654 -6
maks commented 2026-05-02 22:19:46 +02:00
Owner
Copy link

When the user toggles karaoke mode on a Navidrome track, the backend
downloads the track, runs htdemucs (4-stem) via Demucs to separate
vocals, sums the non-vocal stems into an instrumental WAV, and caches
it under data/stems/. The Intel N150 iGPU (torch.xpu) is used when
available, falling back to CPU.

Three new API endpoints: GET/POST /api/karaoke/{id}/status|prepare|stream.
The frontend polls every 3s while processing, then seamlessly reloads
the track at the current playback position with the karaoke stream.
When karaoke first becomes ready, the NowPlaying panel auto-opens so
lyrics are visible for a full karaoke experience.

Co-Authored-By: Claude Sonnet 4.6 noreply@anthropic.com

When the user toggles karaoke mode on a Navidrome track, the backend downloads the track, runs htdemucs (4-stem) via Demucs to separate vocals, sums the non-vocal stems into an instrumental WAV, and caches it under data/stems/. The Intel N150 iGPU (torch.xpu) is used when available, falling back to CPU. Three new API endpoints: GET/POST /api/karaoke/{id}/status|prepare|stream. The frontend polls every 3s while processing, then seamlessly reloads the track at the current playback position with the karaoke stream. When karaoke first becomes ready, the NowPlaying panel auto-opens so lyrics are visible for a full karaoke experience. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add karaoke mode with on-demand vocal separation via Demucs
Some checks failed
CI / Backend (lint + type-check + test) (pull_request) Failing after 1m52s
Details
CI / Frontend (lint + check + test) (pull_request) Has been cancelled
Details
a91cfd3d00 
When the user toggles karaoke mode on a Navidrome track, the backend
downloads the track, runs htdemucs (4-stem) via Demucs to separate
vocals, sums the non-vocal stems into an instrumental WAV, and caches
it under data/stems/. The Intel N150 iGPU (torch.xpu) is used when
available, falling back to CPU.

Three new API endpoints: GET/POST /api/karaoke/{id}/status|prepare|stream.
The frontend polls every 3s while processing, then seamlessly reloads
the track at the current playback position with the karaoke stream.
When karaoke first becomes ready, the NowPlaying panel auto-opens so
lyrics are visible for a full karaoke experience.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Fix audio loading: use demucs AudioFile instead of torchaudio.load
Some checks failed
CI / Backend (lint + type-check + test) (pull_request) Failing after 1m50s
Details
CI / Frontend (lint + check + test) (pull_request) Successful in 3m1s
Details
f8627c18da 
torchaudio 2.x switched its default backend to torchcodec which is not
installed. demucs' own AudioFile uses ffmpeg internally and handles all
formats without extra dependencies.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Some checks failed
CI / Backend (lint + type-check + test) (pull_request) Failing after 1m50s
Details
CI / Frontend (lint + check + test) (pull_request) Successful in 3m1s
Details
This pull request has changes conflicting with the target branch.
  • backend/sonus/api/__init__.py
  • backend/sonus/config.py
  • frontend/src/lib/api.ts
  • frontend/src/lib/audio/player.ts
View command line instructions

Manual merge helper

Use this merge commit message when completing the merge manually.

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u origin feature/karaoke-mode:feature/karaoke-mode
git switch feature/karaoke-mode

Merge

Merge the changes and update on Forgejo.

Warning: The "Autodetect manual merge" setting is not enabled for this repository, you will have to mark this pull request as manually merged afterwards.

git switch master
git merge --no-ff feature/karaoke-mode
git switch feature/karaoke-mode
git rebase master
git switch master
git merge --ff-only feature/karaoke-mode
git switch feature/karaoke-mode
git rebase master
git switch master
git merge --no-ff feature/karaoke-mode
git switch master
git merge --squash feature/karaoke-mode
git switch master
git merge --ff-only feature/karaoke-mode
git switch master
git merge feature/karaoke-mode
git push origin master
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
maks/sonus!2
No description provided.