Wan 2.7 Spicy i2v: Add Your Own Audio Track to NSFW Video

Key Takeaways

• The audio field on alibaba/wan-2.7-i2v-spicy takes a URL to a track you already have and syncs it to the video the model generates — that's upload-and-sync, not generation.
• The same endpoint also exposes audio_enable, the model-generated-audio flag used on Wan 3.0 Spicy and Seedance 2.5 Spicy elsewhere. They're two separate mechanisms on one model — pick one per request.
• Siray's spec confirms audio accepts an audio URL and nothing more; exact format, duration-alignment behavior, and file-size limits are not published. Test with your own file before building a production pipeline around it.
• Among the Spicy-tier i2v endpoints checked here, only Wan 2.7 Spicy accepts an uploaded track. Wan 3.0 i2v Spicy has no audio field at all — audio_enable is its only audio option.

Two things get called "audio" on Siray's Spicy-tier video models, and they don't do the same job. Wan 3.0 Spicy and Seedance 2.5 Spicy use audio_enable, a flag that tells the model to compose synchronized sound on its own — no file required. alibaba/wan-2.7-i2v-spicy works the other way: its audio field takes a URL to a track you already have, and the model syncs that track to the clip it animates from your still image. This piece is about the second one.

One Endpoint, Two Audio Mechanisms

Siray's live spec for alibaba/wan-2.7-i2v-spicy lists both fields on the same request body: audio (string, "Input audio URL") and audio_enable (boolean, "Whether to generate audio along with video"). Our overview of this endpoint covers audio in a single spec-table line and doesn't mention audio_enable at all — worth flagging here so the two don't get conflated. Set audio when you already have a voice line, a soundtrack, or a sound effect you want the clip built around. Set audio_enable when you'd rather describe the sound in your prompt and let the model generate it from scratch. Sending both in one request isn't a documented use case — treat them as alternatives, not a combo.

Compare that to alibaba/wan-3.0-i2v-spicy: its spec has audio_enable but no audio field at all. If uploading your own track matters to your workflow, Wan 2.7 Spicy i2v is currently the endpoint that supports it — Wan 3.0 Spicy's native-audio path, covered on its own overview, doesn't take an uploaded file.

What audio Actually Does — and Doesn't Say

The field takes a URL and nothing else is documented: no accepted codecs, no stated sample-rate requirement, no published behavior for what happens when the track runs longer or shorter than the generated clip, and no file-size ceiling. That's not a gap in this guide — it's the current state of the public spec. If your workflow depends on exact trim or loop behavior at the clip boundary, test it directly against a short track and a long one before committing a production pipeline to it.

duration on this endpoint is capped lower than some siblings: 2-15 seconds, versus Wan 3.0 Spicy's 2-30. Match your track length to that ceiling before you upload it — nothing in the spec suggests the model extends a clip to fit a longer track.

Quick Start

{  
"model": "alibaba/wan-2.7-i2v-spicy",  
"prompt": "...",  
"image": "...",  
"audio": "https://example.com/your-track.mp3",  
"duration": 8,  
"size": "1080p"
}

model, prompt, image, duration, and size are required; audio is additive. Drop it and add audio_enable: true instead if you want the model to generate the sound rather than sync one you supply.

Where This Fits

For NSFW work specifically, an uploaded track is the option that keeps a specific voice performance, a licensed sound effect, or a consistent line of dialogue across multiple generated clips — something audio_enable's from-scratch generation can't guarantee take to take. It's the more deliberate choice when the audio itself is part of the creative direction rather than an afterthought.

Pair this with the NSFW video model roundup if you're still deciding which endpoint fits a given clip, or with the Wan 2.7 Spicy i2v overview for the rest of the endpoint's parameters — resolution, negative prompts, and image-input handling.

Create your free Siray account and sync your own audio to a Wan 2.7 Spicy i2v generation now.

Siray provides access to uncensored AI generation models strictly for legal adult creative use. All content must comply with applicable laws. Siray enforces a zero-tolerance policy on CSAM and any content involving minors. Uncensored access means no excessive content filtering beyond legal requirements — it does not mean illegal content is permitted.