Models

Research model for free-form audio synthesis and transformation from text and optional audio.

Full-song music-generation model in Pika's audio-model family.

琴乐音乐大模型

TME music model applied to AI song creation.

琴韵音乐大模型

TME music model applied to AI singing and voice workflows.

Realtime Audio Variational autoEncoder for fast neural waveform synthesis and timbre transfer.

Real-time text-to-speech API operated by Respeecher.

리마 AI

Emotionwave engine for music analysis, generation, recommendation, and automated performance systems.

ByteDance suite for controllable song generation, score-to-audio editing and zero-shot singing voice conversion.

LeVo 2 open checkpoint for multilingual song generation from text and audio prompts.

Deezer pretrained TensorFlow separation model producing 2 stems.

Deezer pretrained TensorFlow separation model producing 4 stems.

Deezer pretrained TensorFlow separation model producing 5 stems.

Lightweight polyphonic audio-to-MIDI model with pitch-bend detection.

Asynchronous hosted API for Stable Audio 3 text-to-audio, audio-to-audio and inpainting.

Generative-audio model family with commercial and open-weight variants.

Stable Audio 3 exact tier for low-latency high-volume generation, text-to-audio, audio-to-audio, inpainting.