Seed Audio 1.0

### TL;DR

Seed Audio 1.0 is a unified, multimodal audio generation model designed to create complete audio scenes, including speech, music, ambient sound, and sound effects from a single prompt. It enables creators to direct complex audio environments with fine-grained timing control and voice consistency without the need for manual post-production stitching.

Key Insights & Metrics

Key Features

• Full-scene audio orchestration of dialogue, music, and ambient sound
• Zero-shot voice cloning from up to three short reference clips
• Multi-character dialogue generation with distinct emotional and stylistic delivery
• Prompt-level timing control with 100ms precision
• Long-form audio generation up to 2 minutes per pass with continuation support
• Cross-lingual synthesis supporting over 20 languages including English and Chinese

Discussion