Human + source
~20 fields as a baseline
Annotations supplied with the content. Exported fields vary by content type and catalog.
Give audio more context.
Musical, descriptive and technical fields for your workflow, with human annotations and machine enrichment identified separately.
- Tempo
- 129 BPM
- Key
- D minor
- Genres
- Hip Hop/Rap / Rock
- Mood
- confident / energetic
01 / Provenance
Keep the annotation and its source together.
A field’s origin helps you decide how to use it. Keep source annotations and machine-derived attributes identifiable in the same schema.
Machine enrichment
up to ~400 attributes
Automated attributes, available or added after ingestion. Coverage varies; these are machine-derived estimates, not human ground truth.
Descriptor vocabulary: 10,000+ tags. Subgenre taxonomy: ~3,000 subgenres. These describe the vocabulary and taxonomy, not the number of tags attached to every recording.
02 / Field families
From musical structure to technical detail.
Available fields can support selection, search, classification and model conditioning. The field set and coverage are defined for each dataset.
- Identity + relationships
- Track ID, track type, audio filename, associated stems.
- Join recordings, component assets and their descriptive records.
- Musical structure
- BPM and BPM range, perceived tempo, key, mode, time signature, genre and subgenre.
- Define musical subsets and conditioning labels.
- Meaning + emotion
- Descriptions, keywords, augmented keywords, mood, emotional profile and dynamics, character, movement, valence, arousal, energy and brand values.
- Organize semantic search and classification labels.
- Instruments + voice
- Instrumentation, instrument presence, language, vocal style, vocal gender, vocal presence and voice-over.
- Select sound and voice characteristics. Instrument labels do not establish stem availability.
- Lyrics
- Transcription, theme and explicitness.
- Connect lyrical content and text where available.
- Production + segmentation
- Typical purpose, production style, musical era and musical segmentation.
- Add production context and available musical boundaries.
- Technical audio
- Bit depth, channels, duration, stereo/mono and sample rate.
- Check file requirements separately from semantic annotations.
03 / A usable schema
Consistent values. Connected records.
Normalize the vocabulary
Use consistent labels. Demo 1 displays its source key as D minor.
Keep relationships explicit
Track IDs, files and associated stems connect the asset records. Technical properties remain distinct from descriptive annotations.
Deliver in CSV / JSON
Master CSV and/or JSON records accompany the agreed content. Confirm required fields, provenance and missing-value handling as part of the brief.
Work with ERISV
Which fields does your workflow need?
Bring the labels, source requirements and output structure you want. We’ll confirm available coverage and any selected enrichment.