iZotope Introduces Stem Separation Software Photon
· re:union Radio editorial · 1 min read
iZotope has launched Photon. The tool utilizes machine-learning models running on local computer GPUs for stem splitting and audio repair.
iZotope has introduced Photon, a dedicated software solution for stem separation. This release marks the company's first software launch following its acquisition by Boris FX in July, which occurred after the sale of former parent company Native Instruments to InMusic. The models were created in collaboration with the Boris FX team.
Overview of Features and Models
Photon operates as a standalone application or as an ARA plugin inside compatible DAWs, enabling offline processing on the project timeline. All tasks run locally on the computer's GPU. The software incorporates twelve custom-trained machine learning models:
- A music separation model splits mixed tracks into Voice, Bass, Drums, and Other.
- A drum separation model isolates Kick, Snare, Hi-hat, Toms, and Cymbals.
- A voice model divides files into Voice and Music stems for remixing or sampling.
- A Cinema model splits audio and video tracks into dialogue, music, and sound effects.
- Voice Restore and ambient models clean up noise across indoor and outdoor voice recordings.
Interface and File Export
The interface presents audio as a layered waveform, rendering each stem in an individual color. Users can preview material, adjust volume levels, and engage solo or mute buttons before saving. An integrated Model Manager allows downloading new or updated models. Stems can be exported in WAV, FLAC, and MP3 formats.
Source: MusicRadar
