Music selected for your model.
Professional instrumental and vocal recordings, selected for the genres, languages and musical characteristics your model needs.
Available inventory
01 / Selection
Start with the music, then define the set.
Choose from the available inventory, with assets and metadata coverage confirmed for your selection.
- Instrumental titles
- 12M
- Vocal titles
- 1M
- Languages
- 50+
- Curated catalog sources
- More than 40
Choose the characteristics that matter.
Specify instrumental or vocal content, genre, language, instrumentation, tempo, mood and energy alongside the scale and metadata you need.
These inventory figures span the catalog network. Coverage varies by catalog.
02 / Asset package
A recording can connect to more.
Keep the finished audio, available components and descriptive records linked, while making their coverage explicit.
Master recordings
Professional instrumental and vocal source audio, with technical specifications agreed for the delivery.
Stems + MIDI
Original multitracks exist in select catalogs; other stem assets are separated or derived. Original MIDI is available for select content.
Explore stem typesMetadata
Human/source annotations and selected machine enrichment can accompany the audio in structured records.
Inspect metadata03 / Source + delivery
Professional source audio. Usable records.
- Source formats
- WAV / FLAC
- Typical sample rates
- 44.1 / 48 kHz
- Typical bit depths
- 16 / 24-bit
- Channels
- Generally stereo
- Metadata
- Master CSV and/or JSON
Inspect a representative asset.
Follow one master recording through its supplied stems and source metadata. Coverage in this example is specific to that recording.
Open the Data LabSource specifications describe the licensable files. Browser audio previews use separate delivery encodings.
Work with ERISV
What should your music corpus contain?
Share the musical characteristics, asset types and scale you need. We’ll discuss an available selection and the applicable license.