Audio and Speech
14 August 2026
Pika launches four lower-cost generative audio models
N
New-ZZZ desk
X @pika_labs · 4 days ago
Pika Labs has introduced four generative audio foundation models covering soundtracks, music, sound effects, and speech. Pika Soundtrack can create audio from video, while the other models target more specialized forms of sound generation.
The company says its improvements in model training and inference have made the tools up to 20 times cheaper than competing audio models. It also claims Pika Soundtrack is twice as cost-efficient as Hunyuan Foley, which it describes as the only comparable video-to-audio model.
Pika Soundtrack, Pika Music, Pika SFX, and Pika Speech are initially available exclusively through the Pika API Club.
Why it matters
- —Lower generation costs could make advanced audio tools accessible to more developers and media creators.
- —The model family brings video-to-audio, music, sound effects, and speech generation under one API offering.
- —More efficient training and inference could intensify price competition in generative media.
Key facts
- Pika Audio consists of four foundation models for soundtracks, music, sound effects, and speech.
- Pika claims the models can cost up to 20 times less than other audio models on the market.
- The company says Pika Soundtrack is twice as cost-efficient as Hunyuan Foley.
- The models are initially available exclusively through the Pika API Club.
The full text is in the original source. Here we provide a brief summary and key facts.
