Seedance 2.5 Audio Reference: 10 Files, 30 Seconds, and the Field Kling 4.0 Never Lists
Seedance 2.5 accepts up to 10 reference audio files, 30 seconds combined, and switches audio generation on by default. Kling 4.0's input schema on the same platform has no audio field of any kind, in 4.0 or in either older version — read from both model schemas on 27 October 2026.
- Model
- Seedance 2.5
- Observed
Seedance 2.5 takes up to 10 reference audio files, 30 seconds combined, and generates a soundtrack by default. Both facts are in the model's input schema, not in a comparison video. Kling 4.0's input schema, on the same platform, lists no audio field at all — no audio input, no audio switch. Neither does Kling 2.5 or 2.1, which is the part that decides what this finding can and cannot mean.
"Supports audio" turns out to be three separate questions that get collapsed into one adjective. Can the model take audio in? Does it put audio out? Can you turn the audio off? Seedance 2.5 answers yes, yes, yes. Kling 4.0's published inputs answer none of them, because the question never appears.
Everything below comes from two kinds of page, both read on 27 October 2026 without a login and without spending a credit: the vendors' own product pages, and the input schemas embedded in the model pages on Replicate, where both models are published under the vendors' own accounts. No clip was generated for this page. That limit is stated again at the end, where it matters.
The three audio questions, separated
A model that writes a soundtrack and a model that listens to one are doing opposite things with the same word. The comparison videos that rank these models measure the first thing — whether the sound is convincing — and stop there. The second thing, whether you can hand the model a voice and have it follow that voice, is a different capability with different failure modes, and it is the one that decides whether a workflow is possible at all.
There is a third question underneath both: whether the audio can be switched off. A model that always renders sound is not the same product as one that renders it on request, and the difference only shows up when you are cutting a clip into a track you already have.
What Seedance 2.5 actually accepts
BytePlus describes the model with a number and a list: "30s narratives, 50 multimodal inputs, and precision editing in one model," and separately, "Bring together images, video, and audio references so the model can follow complex character setups, product details, scene structures, camera moves, and brand assets." The same page promises that it "seamlessly generates synchronized audio and video with millisecond precision across all sonic layers, from ambient effects to human speech."
That is a vendor claim about capability. The field list is where it becomes something you can act on. Seedance 2.5's schema on Replicate defines three reference arrays, each with its own ceiling:
| Field | Ceiling | What the schema says it is for |
|---|---|---|
reference_images | up to 30 | "character consistency, style guidance, and scene composition" |
reference_videos | up to 10, 30s combined | "motion transfer, style reference, editing, and extension" |
reference_audios | up to 10, 30s combined | "audio-driven generation and lip-sync" |
Two constraints sit inside those descriptions and are easy to miss until a request fails. Audio references cannot stand alone — the schema requires "at least one reference image or video" alongside them. And the reference arrays are mutually exclusive with first-frame and last-frame input: a clip that starts from image or last_frame_image cannot also carry references. So the 50-input budget is not a bigger version of image-to-video; it is a different mode.
Audio output is a separate switch. generate_audio is a boolean, default true, described as generating "synchronized audio with the video, including dialogue (use double quotes in prompt), sound effects, and background music." Default-on is worth noting for anyone who assumed silence was the resting state.
The three ceilings add up to 30 + 10 + 10 = 50, which matches the "50 multimodal inputs" on the product page. That addition is this site's arithmetic, not a vendor statement — BytePlus never says how the 50 are divided, and the match is recorded as a coincidence that happens to be exact, not as a published quota.
What Kling 4.0 lists instead
Kling's own product page is where the comparison gets strange. Its Kling 4.0 block is headed "Audio-Visual Performance for Realistic Results" and reads, in full: "Stable Dynamic Motions and an immersive audio-visual experience, taking video generation to professional-grade production." The word audio appears. What audio means there is never specified — not whether it is generated, not whether it can be supplied, not whether it can be turned off.
The schema is more decisive, because a schema cannot be vague. Kling 4.0 on Replicate defines these inputs:
| Field | Ceiling | Purpose |
|---|---|---|
start_image | — | first frame, jpg/png, 30MB max |
middle_images | — | keyframes between first and last |
end_image | — | last frame |
reference_images | 10 total across all image inputs | "characters, objects, scenes or style" |
reference_videos | — | motion, camera work, style; 3–30s each |
base_video | — | video to edit; output inherits its length and ratio |
There is no audio row. Not an audio input, not a generate_audio flag, not a mute parameter. The string "audio" appears exactly once on that page and it belongs to a platform feature toggle in a permissions object, not to the model.
The comparison that matters is not Kling against Seedance. It is Kling against Kling. The same account publishes Kling 2.5 Turbo Pro and Kling 2.1 on the same platform, and neither of those schemas has an audio field either. Which means this page cannot say "Kling 4.0 lacks audio input" — the honest reading is that on this channel, three Kling versions in a row expose no audio field, and the reason could be the model, the platform's integration, or a capability the vendor keeps behind its own API. Kling's own API documentation would settle it; that documentation renders entirely in JavaScript and returned an empty shell on every attempt here, so it is recorded as not read.
The complaints that now have a mechanism
Two comments under the comparison videos describe symptoms that the field lists explain, at least partly.
"Kling 4 doesn't support audio reference (for character voices) on any platform I used it on, which is a bummer." — @@vfxfan, YouTube comment, 2 likes, read 27 October 2026
"I also find Kling 4.0 very plastic. But the worst was that it adds music even if you write no music. So for cinematic workflow is not good." — @@PyrosAICreative, YouTube comment, 1 like, same date
The first is a user's survey of the platforms they happened to try; this page's schema reading agrees with it on one platform and is not evidence about the others. The second is a complaint about output, and the schema shows a plausible route: with no generate_audio field, there is nothing in the request that could carry a "no music" instruction except the prompt itself. That is a mechanism worth testing, not a proven cause — one generation with no music, silent in the prompt would show whether the audio track survives it, and that generation was not run here.
What this does not settle
No clip was generated for this page. Whether Seedance 2.5's soundtrack is any good, whether Kling 4.0 adds music when told not to, and whether either model's audio track is present at all in a delivered file are all unmeasured here. Reading a schema tells you what a request can contain; it does not tell you what comes back. Anyone quoting this page for audio quality is quoting something it does not say.
The schema readings are platform-side. Replicate publishes both models under the vendors' own accounts, which is why the two sides are comparable at all, but a field list on a hosting platform is an integration, not a specification. Kling's native API may well accept audio; that documentation could not be read. Conversely, Seedance 2.5's 10-audio ceiling is this channel's ceiling — a direct BytePlus integration could differ.
The 50-input arithmetic is this site's, and the match with the vendor's number is not proof of a 30/10/10 allocation. It is a sum of three published ceilings that lands on a published total.
And the reference limits are ceilings, not recommendations. Thirty images plus ten videos plus ten audios is a legal request, not a tested one — no request at that size was sent, so nothing here describes what happens at the boundary.
Prices, tiers and model versions on both sides move without announcement. The readings behind this page are dated 27 October 2026 and carry the channel each was taken from; re-read before quoting.
Related on this site: Kling 4.0 vs Seedance 2.5: what five seconds actually costs — the money side of the same pair, with the tier arithmetic. Seedance 2.5 daily generation cap — how many clips the free allowance actually produces. Video to video AI free — video as a motion reference, which is the neighbouring input type.