.png)
I’ve closely followed the rise of AI music generation—especially as it complements our work in visual AI. When Stepfun AI and Ace Studio launched Ace-Tep, I immediately saw its potential. This fully open-source, Apache 2.0–licensed music model is a game-changer for anyone working with creative AI.
A Closer Look at Ace-Tep’s Capabilities 🧠
Ace-Tep is a 3.5B parameter model capable of generating 4-minute songs in ~20 seconds on an A100 GPU. In my hands-on experience, it rivals the output of Suno AI, Udio AI, and Refusion—but with one key difference: it’s completely open source.
Key features include:
- 🎤 Lyric generation + vocal synthesis
- 🌍 Support for 19 languages (especially strong in English, Chinese, and Russian)
- 🎻 Multiple genres and instruments
- 🧩 Audio inpainting for precise edits
- ✏️ Editable lyrics using flowchart-based workflows
- 🧱 Lora adapters for lyrics-to-vocal and text-to-sample workflows
This combination makes Ace-Tep one of the most versatile and customizable music generation models currently available.
Running Ace-Tep: From Cloud to Local ⚙️
Ace-Tep can be tested via a free Hugging Face demo, but serious users will want to run it locally.
💻 Hardware requirements (approx. 20GB VRAM recommended):
- NVIDIA 3090 → ~5 seconds per minute of audio
- NVIDIA 4090 → ~2 seconds per minute
- Apple M2 Max → Functional, but slow
Despite these demands, it’s impressively fast—generating entire tracks in under 30 seconds on top-tier consumer GPUs.
🌟 As open-source adoption grows, we expect simplified one-click installers and cloud-hybrid options to appear soon.
🎼 Why Ace-Tep Matters to Creative AI
At Promptus, we’re excited about Ace-Tep’s potential to blend audio and visual AI workflows. It fits right into our Cosyflows system, where users orchestrate multimedia workflows with drag-and-drop simplicity.
Coming soon from the developers:
- 🎤 Rap Machine — fine-tuned for high-flow lyricism
- 🎚️ Stem Gem — export individual instrument stems
- 🎶 Singing to Accompaniment — turn vocals into full backing tracks
💡 For creators, this means original soundtracks for videos, adaptive music for games, or collaborative inspiration for musicians—all within one creative pipeline.
My Hands-On Take 🎧
After spending hours with Ace-Tep, here’s what stood out:
✅ Melodic and country genres sound great
🛠️ Advanced settings allow deep customization (inference steps, guidance scale, etc.)
🔁 You can regenerate specific sections or extend compositions organically
📥 Upload and modify existing audio files—a rare feature in commercial tools
It’s not perfect yet (rap delivery needs refinement), but future fine-tunes are already in the works.
Looking Ahead: A New Era of AI Music Creation 🔮
As we expand Promptus Studio to cover more aspects of creative work, tools like Ace-Tep represent the next frontier: open, powerful, and creator-first.
✅ No API fees
✅ No usage limits
✅ Full creative control
Whether you’re using Promptus Web or Promptus App, you’ll soon be able to plug in audio generation to your no-code AI workflow.
Explore the Future of Creativity 🚀
Ready to experience what’s possible when music and visual AI come together?
👉 Visit promptus.ai to get started.
Ace-Tep’s emergence is a powerful reminder that open-source AI is not just catching up—it’s leading. And at Promptus, we’re building the bridge between your imagination and tools that can finally keep up.
Let’s create something unforgettable.
.avif)
