Pika Unveils Four Generative Audio Models
Pika introduced four generative audio models for synchronized video soundtracks, music, sound effects, and speech, available through the Pika API Club.
Development Pika Labs
Published 2026-08-14 (Pika API Club)
Four Audio Models
Pika has unveiled a product family of proprietary audio models covering generative sound more broadly. The family consists of four models: Pika Soundtrack, which adds sound to video; Pika Music, which creates music; Pika SFX, which generates sound effects; and Pika Speech, which synthesizes speech.
Pika Soundtrack interprets movement in video and synchronizes sound effects, music, ambient sound, and speech to fit a scene. Pika Music can create tracks up to six minutes long by combining text, lyrics, vocal references, and music references.
Pika SFX generates up to 20 seconds of 44.1kHz stereo sound effects from natural-language instructions. Pika Speech creates 48kHz speech using preset voices or short reference recordings, and users can also specify delivery characteristics such as intonation and speaking speed.
The four models are available through Pika API Club. Pika said it has lowered costs compared with the models used for comparison through efficient training and inference; the specific date for the pricing comparison is August 14, 2026.