V20 · Audio / Text / Image / URL · Reference-based review
The first interaction is always the input module: upload audio, paste a URL, paste text, upload image, or add references. Spectrograms and analytical graphics appear below the intake area, never before the user understands what to add.
Upload-based analysis + professional visual console.
Platform classification with metadata-aware fallback.
Reference-based writing profile signals.
Synthetic visual-risk markers and preview.
Upload audio for local browser-based review. This module comes first; all charts and interpretation follow below.
Add local AI/human references for comparative calibration.
Use low text influence for serious review. Text analysis is advisory and never proof of authorship.
Required measurement layer: dynamics, repetition, waveform behavior, timbral constancy and temporal change.
requiredalways-onCompares the target file against imported AI-generated reference examples.
reference audioblueCompares the target file against known human-made material to reduce false positives.
human contrastblueCompares features against abstract provider profiles. Useful as a weak signal, not proof.
heuristicamberReduces false positives when constant timbre or repeated articulations may come from sample libraries.
Kontaktvirtual instrumentsEstimates voice-related synthetic signals when vocal material is marked.
voiceadvisoryGreen cards are verified signal categories. Amber cards are heuristic profile models. These are contextual signals, not proof of origin.
Publicly documented SynthID watermark coverage path. Use official verification where available.
verified signalVoice-centric profile for AI singing and synthetic vocal production patterns.
heuristic profileFull-song generation profile. Best used with imported Suno reference audio samples.
heuristic profileSong-generation profile with stronger arrangement movement. Best used with imported Udio reference audio.
heuristic profileReference category for compatible external audio watermark/detector workflows.
verified technologyPaste a media URL first. Auralit classifies the source and uses metadata-aware intake where direct stream analysis is restricted.
Paste text first. The analysis compares writing-pattern signals against human, synthetic and hybrid reference profiles.
No text analyzed.
Upload image first. The visual console then provides synthetic-risk context and reference comparison.
The center node is the current submission. Surrounding clusters represent human, hybrid and synthetic references.
Auralit does not provide definitive authorship proof. It provides explainable comparative indicators, reference-based similarity signals and synthetic-risk context for professional review.