Soundtrack, Music, SFX and Speech cover video soundtrack, song, sound effects and synthetic voice, with generation times much lower than the time listened to.
On August 14, 2026, Pika released four audio generation models at once, initially reserved for members of its Club API. Pika Soundtrack starts with a video and produces the synchronized soundtrack — music, speech, moods and effects tuned to movement. Pika Music composes complete tracks of up to six minutes from text, lyrics, a voice or a reference track. Pika SFX creates stereo sound effects of up to twenty seconds from a natural language description. Pika Speech, finally, does expressive speech synthesis in 48 kHz, with pre-recorded voices or cloning.
The argument put forward is speed. A ninety-second piece is announced in 6.21 seconds of calculation, nearly fifteen times faster than its listening time; a sound effect comes out in less than a second; speech displays a real-time factor of 0.02, a minute of voice requiring about one second of generation.
Pika claims the lowest rates on the market, with a cost advantage of up to twenty times according to comparisons based on public prices as of August 2026. The bet is clear: in a segment where competitors charge for the second generated, make audio a commodity cheap enough to be called en masse by automated video production lines.
Read the full article on the original source
Go to the source →