LIVE
Google Introduces "Gemini 4 Carbon" as Next Step in Its Model Lineup28/10/26 · Google|DeepMind unveils Genie 4, a new generation of world model27/10/26 · Google DeepMind|Anthropic files IPO prospectus, with risk disclosures taking center stage26/10/26 · Anthropic|Robot policy trained only in simulation matches top real-world systems26/10/26|OpenAI unveils GPT-6.1 Terra, a model built for computer use26/10/26 · OpenAI|Google's AI gives evasive answers about who won the 2020 US election28/10/26 · Google|NVIDIA introduces Nemotron 3.5 Transcribe, a promptable multilingual transcription model28/10/26 · NVIDIA|OpenAI and ServiceNow deepen partnership on enterprise workflows28/10/26 · OpenAI|Failed startups' data finds a second life training AI models28/10/26|Anthropic expands AI for Science program to 20 foundations and research institutions28/10/26 · Anthropic|Anthropic shares early results on using Claude to accelerate drug development27/10/26 · Anthropic|GEPA-Curator: Filtering Failures to Better Train LLM Agents27/10/26|Google Introduces "Gemini 4 Carbon" as Next Step in Its Model Lineup28/10/26 · Google|DeepMind unveils Genie 4, a new generation of world model27/10/26 · Google DeepMind|Anthropic files IPO prospectus, with risk disclosures taking center stage26/10/26 · Anthropic|Robot policy trained only in simulation matches top real-world systems26/10/26|OpenAI unveils GPT-6.1 Terra, a model built for computer use26/10/26 · OpenAI|Google's AI gives evasive answers about who won the 2020 US election28/10/26 · Google|NVIDIA introduces Nemotron 3.5 Transcribe, a promptable multilingual transcription model28/10/26 · NVIDIA|OpenAI and ServiceNow deepen partnership on enterprise workflows28/10/26 · OpenAI|Failed startups' data finds a second life training AI models28/10/26|Anthropic expands AI for Science program to 20 foundations and research institutions28/10/26 · Anthropic|Anthropic shares early results on using Claude to accelerate drug development27/10/26 · Anthropic|GEPA-Curator: Filtering Failures to Better Train LLM Agents27/10/26|
AppRisingofficial publisher

YuE

App : A product you use directly

Generate complete songs with lyrics, locally, open source.

Powerful but demanding: 24 GB of VRAM recommended.

Official site GitHub · 12,679 ★
Overall score
7.1 /10
Security: one caveat, see the card
Utility
7.5
Reliability
10.0
Security
One caveat
Ease
Technical · Technical

Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.

What it does, concretely

Creating a complete song with melody, chords, vocals and accompaniment normally requires advanced musical skills and hours of work. YuE2 makes it possible to generate a full song from lyrics and a style description, entirely on your own machine.

The process relies on a multi-stage architecture: YuE2 first produces a symbolic score (melody and chords in ABC format) that you can read and edit before launching audio generation. This score is then converted into semantic tokens, then acoustic latents, and finally decoded into 48 kHz stereo audio. The Python API exposes each stage separately through plan(), generate_semantic(), synthesize() and decode().

YuE2 is aimed at musicians, developers and creators who want to compose, reinterpret existing tracks or iteratively edit compositions without relying on an online service. The ability to inspect and modify the score before the final render is a concrete advantage for anyone who wants to keep control over the musical structure.

However, the tool is not suitable for users without an NVIDIA GPU with at least 24 GB of VRAM, or for those looking for a simple graphical interface: installation and use require the command line and Python. Model weights are released under a CC BY-NC 4.0 licence, which excludes any commercial use.

What it offers

  • ·Full song generation (vocals and accompaniment) from lyrics and a style prompt.
  • ·Symbolic planning: YuE2 first produces a melody-and-chord score in ABC format, editable before audio rendering.
  • ·Zero-shot covers: reinterpretation of an existing recording in a new style from its transcription.
  • ·Agentic editing: a conversational agent can modify the score (harmony, tempo, arrangement, lyrics) and generate a new version.
  • ·Multi-stage Python API (plan, generate_semantic, synthesize, decode) for fine-grained control over each phase.
  • ·48 kHz stereo FLAC audio output, with the score, tokens and latents saved alongside in the output folder.
  • ·Compatibility with SheetSage2 to transcribe a source recording into an ABC score before reinterpretation.
  • ·Fully local operation: no data is sent to a third-party server.

In practice

Price
Free
Commercial use
Commercial use: unclear
Available on
linux, windows
Licence
Apache-2.0
Runs locally
Yes, 24 GB VRAM recommended
FR interface
No

Security and data

Rating
Security: one caveat, see the card
Your data
Data stays local
Maintenance
Active
Last release
9 September 2026

Links and repository checked on 25 October 2026 · added on 1 January 2025

How to install YuE

1. Clone the repository: git clone https://github.com/multimodal-art-projection/YuE.git 2. Enter the folder: cd YuE 3. Create a Python 3.12 virtual environment: python3.12 -m venv .venv 4. Activate it: source .venv/bin/activate (Linux) or .venv\Scripts\activate (Windows) 5. Install dependencies: python -m pip install --upgrade pip && python -m pip install . 6. Run a first test: python examples/generate.py --output outputs/first-song (model files are downloaded from Hugging Face on first use; make sure your NVIDIA GPU has at least 24 GB of VRAM before starting).
git clone https://github.com/multimodal-art-projection/YuE.git && cd YuE && python3.12 -m venv .venv && source .venv/bin/activate && python -m pip install --upgrade pip && python -m pip install .

In the same category

Whole category Audio, voice & music →