Pex ACR identification capabilities
Pex’s AI-powered technology utilizes advanced digital fingerprinting and matching techniques to identify content with extreme precision and accuracy.
Audio
Identifies identical sound recordings or parts of them
- • Very short samples (1s)
- • Mashups, multi-track mixes, and remixes
- • Speed changes between 50-200% (half or double speed)
- • Pitch changes between +-1200 cents (i.e. a full octave up or down)
- • Recorded in a noisy environment
- • High compression and audio with a low sample rate
- • Rich stems like vocals, piano, and guitars
- • Effects such as echo, reverb, low or high-pass filters, and similar
Melody
Pex melody fingerprinting and matching technologies identify melodies despite significant differences, including the removal of lyrics. This technology is commonly known as cover song identification.
- • Background noise
- • Tempo changes
- • Transpositions
- • Live concerts
- • Short samples
- • Language changes
- • Different timbres and instruments
AI-generated music
Identifies songs as fully AI generated or human created
Our AI Song Detector provides a binary response for whether or not a song is fully AI generated. It also identifies which platform most likely generated the song.
- • Determines if a track is fully AI generated
- • Broad support across major AI music generation platforms, including Suno, Udio, Boomy, ElevenLabs, others
- • Identifies which model generated a song, when the detector has strong attribution confidence
- • Easily integrated via API
- • Provides results in real time
- • More accurate than competitors
Phonetic
Pex phonetic fingerprinting and matching technologies identify identical underlying compositions by matching the speech or lyrics.
- • Leverages Automatic Speech Recognition
- • Relies on vocal expression of lyrics, not extracted text or lyrics metadata
- • Able to identify songs with the same lyrics with very high accuracy
- • Compliments melody or audio matching
- • Matches different voices, genres, pitches, and tempos
- • Language changes
- • Support for all languages
Voice
- • Analyzes the voice, independent of lyrics, spoken words, or any textual information or metadata
- • Recognizes the exact same voice across different recordings, regardless of whether there is one or multiple voices
- • Matches different songs, rap performances, or speech segments performed by the same, even if previously unknown, individual
- • Able to identify any previously registered singer in a given song
- • Can match real and AI-generated voices to help detect AI-generated music
- • Effective across various genres and languages
Video
Identifies identical videos, or parts of them
- • Very short samples (1s)
- • Grayscale distortions
- • Brightness change
- • Added noise (visual)
- • Mild cropping
- • Text overlay
- • Added logos
- • Filters
- • Livestream
