Video dubbing
A scene with sound
Hear voice, music and sound effects brought together for a scene.
The video includes on-screen descriptions of dialogue, music and effects.Seed Audio 1.5 · Coming soon on Popcraft
Seed Audio 1.5. Everything to hear.
Coming soon on Popcraft. Preview scene audio with dialogue, ambience, sound effects and music from text, video and audio references. The release date and pricing for Seed Audio 1.5 on Popcraft have not been announced.
Explore Seed Audio 1.5 ↓Examples
Hear it in context.
Preview demos supplied with the Seed Audio 1.5 showcase. Popcraft generation is coming soon.
Timestamp control
Voices, music, moments
Hear multiple voices, music and sound effects arranged using timestamps in an English-language ad.
What’s different
More ways into audio.
Combine text, video and audio references to shape the result.
| Input / capability | How it works | Audio 1.5 | Audio 1.0 | Conventional TTS |
|---|---|---|---|---|
| Text-to-speech | Choose a voice and supply the words. | Yes | Yes | Yes |
| Voice cloning | Use one audio clip as the voice reference for speech. | Yes | Yes | Yes |
| Text → audio | Describe speech, sound effects and background music. | Yes | Yes | No |
| Text + audio → audio | Add a reference for voice characteristics or emotion. | Yes | Yes | No |
| Text + video → audio | Describe audio that follows the video’s visuals. | Yes | No | No |
| Text + video + audio → audio | Add an audio reference to specify the dubbing voice. | Yes | No | No |
| Text → music | Describe music up to six minutes long. | Yes | No | No |
One model, more possibilities
Give your visuals a voice.
Capabilities and limits on this preview come from the supplied Seed Audio 1.5 showcase. Final Popcraft features and limits will be confirmed at launch.
From a first soundscape to a new language, shape the audio around the story you want to tell.
6 minmaximum audio length per request
6audio references in one request
33supported languages
01 / VIDEO DUBBING
Sound for the scene
Use a video as a reference to generate voice, sound effects and background music that match its visuals.
Describe the exact dialogue, sound effects and music you want; dubbing is not automatic.02 / VIDEO TRANSLATION
Carry the voice further
Translate spoken content while retaining the original voice characteristics, sound effects and background music.
The documented translation pairs are Chinese or English to 30 target languages.03 / STEM SEPARATION
Room to refine
Output vocals, sound effects and background music as independent stems for further editing.
04 / MUSIC GENERATION
Find the soundtrack
Describe music in text, or generate a song with lyrics. Music can be up to six minutes long.
Before the soundtrack
Start with the scene.
The original scene
Compare this video with the enhanced scene in the examples above.
Video workflows
Build the sound. Keep the story.
Dub or improve a scene
Use video alone, or video with an audio reference, for film dubbing or improving low-quality audio.
- Choose the video reference.
- Add an audio reference when you want a specific voice timbre.
- Describe the exact spoken content, sound effects and background music.
Video dubbing supports 30 languages. It requires a text description; it does not dub automatically.
Play the scene example ↓Translate the spoken content
Start with Chinese or English speech. Translate it into one of 30 supported target languages while retaining voice characteristics, sound effects and background music.
Examples include a translation demonstration, an English Whiplash scene and its Japanese counterpart, plus an English Big Bang Theory scene and a Mexican Spanish version.
View translation languages ↓Video translation examples
Hear the same scene differently.
Across languages
A short demonstration of video translation.
Whiplash · English
Original English scene.
Whiplash · Japanese
Japanese translation by Seed Audio 1.5.
The Big Bang Theory · English
Original English scene.
The Big Bang Theory · Mexican Spanish
Mexican Spanish translation by Seed Audio 1.5.
Stem separation
Take the mix apart.
Supply an audio file to separate vocals, effects and background music into independent outputs for post-mixing. The current version supports up to six tracks.
From audio to editable tracks
Each output can be used individually in a mixing workflow.
The 58-second example separates a mixed audio file into narrator, girl, boy, fantasy music, footsteps and door-closing tracks.
Keep the limits distinct
- Up to six output tracks for stem separation.
- Up to six audio references in a generation request.
- Up to six minutes of generated audio per request.
Separate, then compare
One mix. Six parts.
Stem separation in a scene
A demonstration of vocals, sound effects and music as separate outputs.
Listen to the mix and each isolated track.
Original mix
Narrator
Girl
Boy
Fantasy music
Footsteps
Door closing
Music generation
Describe the soundtrack.
Direct the musical choices
Specify genre, mood, vocals, instruments and structure in your music requirements. Generate music from a text description, or a song with lyrics.
Music can be up to six minutes long.
Five music languages
English, Chinese, Japanese, Korean and Cantonese are supported for music generation.
This music-specific list is separate from the model’s broader 33-language support.
Music example
From requirements to music.
Music generation
Explore the music-generation demonstration.
Language support
Know the language scope.
The model supports 33 languages overall. Translation starts from Chinese or English and supports these 30 targets.
- Chinese ·
zh - English ·
en - Japanese ·
ja - Korean ·
ko - German ·
de - French ·
fr - Brazilian Portuguese ·
pt-BR - Thai ·
th - Indonesian Bahasa ·
id - Vietnamese ·
vi - Malay ·
ms - Filipino ·
fil - Italian ·
it - Russian ·
ru - Dutch ·
nl - Polish ·
pl - Turkish ·
tr - Swedish ·
sv - Castilian Spanish ·
es-ES - Mexican Spanish ·
es-MX - Arabic ·
ar - European Portuguese ·
pt-PT - Finnish ·
fi - Danish ·
da - Norwegian ·
no - Czech ·
cs - Hungarian ·
hu - Greek ·
el - Romanian ·
ro - Hindi ·
hi
Beyond the translation list
The overall 33-language list also includes Tamil, Odia and Bengali. The overall Arabic entry specifies a Saudi accent. These broader capabilities do not expand the documented translation pairs.
Precise control
Make every moment count.
Use timestamps to place voice-over, sound effects and background music where the story needs them.
VOICE
Place the spoken moment.
SOUND EFFECTS
Give the action its texture.
MUSIC
Set the mood around the scene.
Explore the ad’s timing and voice direction
Timestamp ranges place speech, effects and music. Each voice also has its own timbre, accent and delivery direction.
- 0.0–49.5s
Dynamic electronic music underscores the full ad.
- 0.0–1.2s
A crisp splash opens the audio.
- 1.9–4.9s
Young boy with a tender-clear timbre, innocent and imaginative character, Central-Asian British English accent; steady, relaxed, light-hearted and childlike commercial delivery: “BYD is the big ship that I saw at the pier.”
- 5.8–9.0s
Young girl with an agile, translucent voice and lively, endearing character, American English accent; snappy rhythm and slightly contradictory commercial delivery: “No, my grandpa said that BYD is the myth from the East.”
- 10.1–13.2s
Young girl with a bright, sweet, soft-toned voice, Chinese-accented American English; cheerful commercial delivery with a hint of amusement: “My mom works at BYD. She says, BYD, Build Your Dreams.”
- 13.4–15.4s
Children’s laughter, then faster, tech-infused music with camera-shutter effects.
- 15.8–17.8s
Young adult woman with a slightly thick, subtly breathy voice and refined, elegant character, French-inflected American English; admiring, composed and articulate commercial delivery: “Simply, an unforgettable model I've ever seen.”
- 19.3–22.1s
Middle-aged explorer with a deep, rugged, mildly husky voice and mature character, General American accent; bold, resolute and expressive commercial delivery: “He's the most reliable companion to cross the Andes with.”
- 23.2–25.2s
Young adult man with a mid-high range and somewhat flat vocal texture, General American accent; a short, excited commercial delivery: “He's the new beast in the arena.”
- 27.3–29.7s
Young adult man with a deep voice and poised business character, General American accent; soft, relaxed commercial delivery with contented gratification: “BYD is my business lounge on the wheels.”
- 29.7–49.5s
Brand music closes the ad with waves, wind and vehicle sounds.
Good to know.
Seed Audio 1.5 is coming soon on Popcraft and is not available for generation here yet. A release date and pricing have not been announced.
The documentation describes text, video and audio references. A request can use up to six audio references.
No. Provide a text description of the exact content, sound effects and background music you want.
The documented video translation pairs start with Chinese or English and translate into 30 supported target languages. The model supports 33 languages overall.
Seed Audio 1.5 · Coming soon on Popcraft
Hear the whole story.
Explore what you can create with Seed Audio 1.5.
Explore the examples ↑Keep exploring
Related models
Eleven v4
ElevenLabs
Eleven v4 and Eleven v4 Turbo on Popcraft: ElevenLabs' most emotive text-to-speech yet. Inline audio tags, 90+ languages, 10,000-character takes, and a Turbo cut built for real-time voice agents.
Music V2.5
Popcraft
Score videos in seconds with Music V2.5 on Popcraft. Generate AI background music, instrumental BGM, and mood-matched tracks from 3 to 200 seconds long.
ElevenLabs SFX
ElevenLabs
Generate custom sound effects from text with ElevenLabs SFX on Popcraft. 0.5–22s clips, tunable prompt influence, and 48 kHz studio-ready audio.
Seedance 2.5
ByteDance
Seedance 2.5 generates up to 30 seconds of video in a single job, with up to 50 reference assets, controllable video editing, high-fidelity temporal extension, any aspect ratio from 0.4 to 2.5, and native prompting in 10+ languages. Live on Popcraft.