New ComfyUI Music Models: Local FLAC Songs with Vocals (Real Workflow)

In this video, I test the newest local AI music models running inside ComfyUI and show how to generate high-resolution FLAC audio with clean vocals, strong lyrics sync, and surprisingly fast generation times. You’ll see exactly which repositories and custom nodes to install, how to place checkpoints/codecs/config files in the correct folders, and how to avoid common “broken workflow” problems when models update. I compare multiple approaches, including MusicGen Melody (generate music from your humming with a microphone), Tango Flux workflows for lyric-driven songs, and other recent audio models that require multi-part installs. I also explain real hardware considerations like VRAM usage, why some models don’t unload from memory, and when restarting ComfyUI is the simplest fix. If you want a practical, production-style setup for local AI music generation (not cloud tools), this walkthrough will help you install everything, run the workflows end-to-end, and export your results in FLAC, WAV, or MP3. My Music Pack with ComfyUI nodes and more: https://github.com/GeekatplayStudio/S... Models: https://huggingface.co/HeartMuLa/Hear... Info: https://github.com/HeartMuLa/heartlib ComfyUI nodes for HeartMuLA: https://github.com/benjiyaya/HeartMuL... Models: https://huggingface.co/stabilityai/st... Models and Info: https://github.com/declare-lab/Tangoflux ⚠️ AFFILIATE DISCLOSURE: We may earn a commission from purchases made through the links below at no extra cost to you. My recommendations: Hitem3d: https://www.hitem3d.ai/?utm_source=af... TripoAI: http://studio.tripo3d.ai?via=geekatplay Meshy 3d AI: www.meshy.ai?via=geekatplay i10X. :https://i10x.ai?fpr=vladimir24 Topaz AI Video and Photo processing: https://topazlabs.com/ref/1514/ My Patreon webpage -   / geekatplay   Tutorials and packs - https://gumroad.com/geekatplay Tutorials website - https://www.geekatplay.com Photography - https://www.chopinephotography.com #ComfyUI #AIMusic #MusicGen #LocalAI #AIAudio #OpenSource #TextToMusic #AIWorkflow #FLAC #GenerativeAI 00:00 High-resolution FLAC music from ComfyUI on a local machine (newest models) 00:32 What this video covers + where to find all resources and links 00:52 Install my latest custom nodes (git clone into ComfyUI custom_nodes) 01:22 Sonic Holiday repo overview + required components and models 01:43 Models we’ll test: Tango Flux, Stability/OpenAI, and a new Facebook release 02:03 Why installation can be confusing + using the included installers (Windows/Linux) 02:26 Where files go: checkpoints, safetensors chunks, codecs, and configs 02:52 Subscribe/like to stay updated when code or models change 03:06 Restart ComfyUI + how to find nodes by searching “sonic” 03:25 MusicGen Melody node: generate music from humming (44kHz, mono/stereo, sizes) 03:57 Microphone setup + duration control + press/hold to record humming 04:16 Save output in multiple formats (MP3/FLAC/WAV) 04:27 Text prompt mode: pick a model and specify a style (example: K-pop) 04:51 Run a quick test + watch generation progress 05:06 GPU memory usage explanation (models staying in VRAM) + cleanup tips 05:32 Recommendation: restart ComfyUI after you pick a workflow you like 05:51 Switch to Tango Flux “Sonic DJ” for lyric-synced song generation 06:14 Style/voice/duration settings + Bark text-to-voice in the workflow 06:58 Speed demo: ~12 seconds generation + key settings (steps, CFG) 07:28 Best-quality surprise model: tricky install + Python glue code to assemble pieces 08:11 Choose genre/mood/vocals + structured lyrics with tags (verse/chorus/intro/outro) 09:05 Two-minute song generation + waveform preview and save options 09:36 VRAM check: models still loaded + why a restart helps before longer runs 10:06 Full run timing: ~2–2.5 minutes + ~22GB VRAM noted 10:38 Play the result + voice quality and overall impressions 10:51 Links in description + star the repo + closing goodbye