Seed Audio 1.5
ByteDance's audio generation model places dialogue, music, ambient sound, and sound effects in the same scene, supporting generations up to 6 minutes long with stem output.
About Seed Audio 1.5
Seed Audio 1.5 is ByteDance's audio generation model, available on the TopMaker platform. It generates spoken dialogue, music, ambient sound, and sound effects within the same scene, and offers more ways to guide the performance and adjust the final edit. You can start from text, guide the voice with reference audio, or build sound around existing video footage. Compared with version 1.0, a single generation can last up to 6 minutes, supports up to 6 audio references, covers 30 languages and regional variants, and expands input combinations to text, text + audio, text + video, and text + audio + video. The longer generation length and stem output enable richer workflows for storytelling, podcasts, and sound design.
Best for
Best use cases
Generate longer spoken dialogue and scoring for stories and podcasts
Use existing video footage as context and build sound around the visuals
Explore dialogue voiceover across languages and regional variants
Sketch layered soundscapes for film and games, align sound effects to specific moments, and keep editing with stems
Features
Longer generation length
Generate audio up to 6 minutes long in a single run (version 1.0 allowed up to 2 minutes).
More audio references
Supports up to 6 audio references (version 1.0 allowed up to 3) to guide the vocal performance.
Multilingual and regional variants
Covers 30 languages and regional variants (version 1.0 covered 20+ languages).
Multiple input combinations
Supports text, text + audio, text + video, and text + audio + video input combinations (version 1.0 supported text, text + audio, and text + image).
Generate around existing footage
Use existing video footage as context and build sound around it.
Stem output
Supports stem output, making it easy to keep adjusting in your edit.
Pros
- Up to 6 minutes per generation, suitable for longer content like podcasts
- Supports up to 6 audio references to guide the voice
- Covers 30 languages and regional variants
- Supports multiple input combinations such as text + video and text + audio + video
- Supports stem output for easier post-editing
Cons
- Seed Audio 1.5 is labeled "Coming soon" on the page and has not officially launched yet
Pricing Plans
Basic
- 6,000 credits per year
- Approximately 1,200 images or 240 videos
- Access to all models
- Private generations
- Watermark-free creation
- Commercial usage rights
Pro
- 30,000 credits per year
- Access to all models
- Private generations
- Watermark-free creation
- Commercial usage rights
* Pricing information is for reference only. Please refer to the latest announcement on Seed Audio 1.5's official website.