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|
AppWatch outofficial publisher

DiffRhythm

App : A product you use directly

Generate complete tracks in seconds, locally.

The fastest of the local generators.

Official site GitHub · 2,353 ★
Overall score
6.1 /10
Security: one caveat, see the card
Utility
7.5
Reliability
3.0
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 complete music with lyrics and arrangements usually takes hours of studio work. DiffRhythm makes it possible to generate a full song, vocals and instruments included, in a matter of seconds, directly on your own machine, without sending your data to a remote server.

The model is based on a latent diffusion architecture. You provide lyrics in LRC format along with either a reference audio file for style or a text description of the desired genre (for example 'Pop Emotional Piano' or 'Indie folk ballad'). DiffRhythm then produces a track up to 4 minutes and 45 seconds long. It is also possible to generate instrumental music without vocals, edit an existing track, or extend it.

The tool is aimed at developers, researchers, and creators who want to experiment with AI music generation on their own hardware, without relying on an online service. A graphics card with at least 8 GB of VRAM is required to run the base model.

DiffRhythm is not suited for users looking for a simple graphical interface or a turnkey service: installation goes through the command line and requires knowledge of Python environments. People without a sufficient GPU or technical experience will find it difficult to get started.

What it offers

  • ·Full song generation with vocals and instruments, up to 4 minutes and 45 seconds.
  • ·Generation guided by a text description of the musical style, without a reference audio file.
  • ·Generation guided by a reference audio file to reproduce a specific sound style.
  • ·Instrumental mode to produce music without vocals.
  • ·Editing and extending an existing track with understanding of its structure.
  • ·Fully local deployment on Linux, Windows, and macOS, without sending data externally.
  • ·Online demo available on Hugging Face Space for testing without installation.
  • ·Docker installation to simplify environment setup.

In practice

Price
Free
Commercial use
Commercial use: unclear
Available on
linux, windows
Licence
Apache-2.0
Runs locally
Yes, 8 GB VRAM recommended
FR interface
No

Security and data

Rating
Security: one caveat, see the card
Your data
Data stays local
Maintenance
Inactive for over six months

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

How to install DiffRhythm

1. Install espeak-ng for your system (apt-get, yum, brew, or the .msi installer on Windows). 2. Clone the repository: git clone https://github.com/ASLP-lab/DiffRhythm.git then enter the folder. 3. Create a Python 3.10 environment with conda or venv and activate it. 4. Install dependencies: pip install -r requirements.txt. 5. On Windows, set the PHONEMIZER_ESPEAK_LIBRARY and PHONEMIZER_ESPEAK_PATH environment variables, then restart your machine. 6. Run inference with the script matching your use case (audio reference or text prompt); add --chunked if your VRAM is limited to 8 GB.
git clone https://github.com/ASLP-lab/DiffRhythm.git && cd DiffRhythm && pip install -r requirements.txt

In the same category

Whole category Audio, voice & music →