As artificial intelligence continues to work its way into the music industry, a new challenge has emerged: reliable identification of AI-generated voices. At Pex, we’re tackling this issue head-on with cutting-edge technology designed to protect artists and rightsholders in the digital age.
Key Points:
- Pex combines Voice ID and Automatic Content Recognition (ACR) to identify AI-generated voices in music.
- Traditional methods like watermarking and artifact detection have proven insufficient.
- Pex technology can identify AI cover songs, vocal swaps, and new songs with AI voice impersonations.
The growing challenge of AI in music
The rise of AI-generated music, particularly songs that mimic famous artists’ voices, presents a significant challenge to the music industry by flooding music services with tracks that likely contain unauthorized use of an artist’s likeness. Additionally, many of the generative AI models used to generate AI songs or AI voices are likely trained on copyrighted content without proper licensing, leaving original artists in both cases without credit or compensation.
At Pex, our mission of “attribution for all” addresses this challenge directly. We’ve developed a unique combination of Voice ID technology and Automatic Content Recognition (ACR) to safeguard intellectual property, including artists’ distinctive voices.
Why traditional methods fall short
Watermarking
While embedding imperceptible markers (AKA “watermarks”) in AI-generated content seems promising, this method is vulnerable to removal or modification attempts. And today, there are no known watermarking techniques that can withstand removal.
Artifact Detection
Identifying AI-specific anomalies in generated content is an approach that consistently lags behind the rapid advancements in AI technology. Detection methods typically require examples of such content to develop effective tools. Consequently, these detection mechanisms inevitably trail behind the creation of AI-generated material.
Pex’s solution: Voice ID + ACR technology
Our approach doesn’t try to answer the question of whether the content is AI generated as much as it is looking to find if any rights are being infringed through matching a recording, composition, lyrics and voice. Pex leverages a powerful combination of technologies to provide a more accurate, scalable, and deception-proof method for identifying problematic AI-generated music:
- Voice ID Technology: Identifies singers by matching biometric voice traits
- Automatic Content Recognition (ACR): Compares content qualities to find matches
- Audio, Melody, and Phonetic Matching: Enhances overall identification accuracy
Key Features of Pex’s Voice ID:
- Singer Matching: Determines if multiple recordings feature the same singer, even across different musical styles or languages
- Singer Identification: Names the specific singer in an audio file using a comprehensive reference database
Real-world applications
Pex technology has successfully identified various types of AI-generated music, demonstrating its versatility and effectiveness. Let’s take a look at a few examples.
Known Songs with AI Vocal Swaps
Pex technology identified Freddie Mercury’s AI-generated voice in a version of Disney Frozen’s “Let It Go”. By identifying this audio as “Let It Go” and separately identifying Freddie’s voice, we were able to determine this song’s use of AI vocals.
AI Call Out My Name
Original song https://www.youtube.com/watch?v=M4ZoCHID9GI
New Songs with AI Voice Impersonations
Creators aren’t just swapping vocals on original songs, they are creating entirely new songs with AI and generating vocals in the style of well-known artists. We identified this AI song using vocals that impersonate popular band Oasis. Oasis’ distinct vocal style is represented in the track, and our advanced technology is able to identify lead singer Liam Gallagher’s voice throughout.
Oasis – Summer of 95 (2024 New Song AI)
Cover Songs with AI-Generated Voices
We also used a combination of our Voice ID technology and ACR to identify this AI cover of Johnny Cash singing “Barbie Girl” by Aqua. The vocal styles and genres could not be more different here, but thanks to the underlying melody and lyrics, we were still able to find this cover. We then identified the voice as Johnny Cash to further determine that this song was created with AI.
Johnny Cash – Barbie Girl (Cover by There I Ruined it) Restoration
Shaping the future of music copyright protection
As AI-generated content continues to grow, Pex’s innovative approach offers a robust solution for copyright protection. By combining Voice ID with ACR technology, we’re setting new standards in identifying and protecting against unauthorized use of artists’ voices. Our Voice ID technology can match singing voices to their identities, even when there are multiple voices in a recording, or when the voice is AI-generated and is effective across various genres and languages.
Core Pex VoiceID capabilities:
- Voice analysis, independent of lyrics, spoken words, or any textual information or metadata
- Voice recognition across different recordings, regardless of whether there is one or multiple voices
- Speech matching for songs, rap performances, or spoken word performed by the same, even if previously unknown, individual
- Singer identification
- Voice matching, real and AI-generated, to help detect AI-generated music
Identify singers and AI-generated voices with Pex
Are you interested in leveraging Voice ID technology for your company or catalog? We invite you to explore how our solutions can help protect your intellectual property in this new age of AI-generated music.
Contact our team to learn how Pex technology can help safeguard your musical assets.
