SONOVOX
Free packGet the dataset
FAQ

Frequently asked questions

About Sonovox

What is Sonovox?
Sonovox is the world's premier vocal dataset for Audio & Voice AI. We provide over 5,000 full-length, raw, unprocessed studio vocal stem packs — totaling more than 20,000 minutes of male & female, multi-genre pop vocal recordings.
Who uses Sonovox?
Our dataset is trusted by AI companies, researchers, audio startups, music tech platforms, and enterprises building voice models, generative audio tools, and next-gen audio applications.

Licensing & rights

Are the vocals royalty-free?
Yes. All recordings are 100% royalty-free, perpetually licensed, and AI-ready.
Do I need additional clearances to use the data commercially?
No. Once purchased, you have full commercial rights under our perpetual license.
Is the dataset ethically sourced?
Absolutely. All recordings are created in-house with contracted vocalists who have granted rights for AI training and commercial use.

Compliance & procurement

Does the dataset contain any data collected in China, Cuba, Iran, North Korea, Syria, Crimea and other regions of Ukraine controlled by the Russian Federation, Belarus, the Russian Federation, or Venezuela?
No. The dataset does not contain any data collected in those regions.
Does the dataset contain any known pirated materials?
No. The dataset does not contain any pirated materials.
Is the license on the SPDX license list (AGPL, SSPL, OSL, Artistic 2.0, APSL, RPSL, RPL, Elastic 2.0, etc.)?
No, it’s not on the SPDX license list. It is a proprietary, custom commercial license agreement exclusive to companies purchasing directly from us at Sonovox.ai only, so it doesn’t fall under any of those open-source categories. You can read the full agreement here.

Dataset details

What exactly is included?
  • 5,000+ vocal stem packs (lead, doubles, harmonies, ad-libs)
  • 20,000+ minutes of audio
  • Multi-genre coverage (pop, R&B, EDM, rock, indie, ballads, etc.)
  • Both male & female voices
  • Delivered in raw, unprocessed WAV format (studio-grade, no FX)
Are the files labeled or organized?
Yes. Each pack is folder-structured with consistent labeling: Gender, BPM, KEY (lead, background vocals, harmonies, ad-libs).
What formats are supported?
Industry-standard .WAV (44.1kHz, 24-bit) — clean, lossless, and ready for AI/ML pipelines.

Usage & applications

What can I use the dataset for?
  • Training voice models and generative AI systems
  • Testing and benchmarking audio pipelines
  • Developing text-to-speech, voice cloning, and audio synthesis models
  • Building music and vocal-based applications
Can I use it in my commercial app?
Yes. The license covers both research and commercial deployments.
Is this suitable for machine learning?
100%. All stems are raw and unprocessed — AI-ready out of the box.

Access & support

How do I access the dataset?
After purchase, you'll receive secure cloud delivery with full dataset download and perpetual access.
How large is the dataset?
600 GB of high-quality audio data.
Do you offer enterprise support?
Yes. Enterprise customers receive dedicated onboarding, integration guidance, and priority support.
How do I contact support?
Email joseph@sonovox.ai or call +1 (778) 513-9246.