Wan 2.7 silent rows need TTS

Silent plate plus voice is a pipeline, not a surprise. Wan 2.7 text-to-video and image-to-video rows in the Versely catalog carry null audio (catalog). If your brief needs speech, plan TTS and lipsync as explicit steps. Do not discover silence after you already promised a talking UGC cut for Friday.

What null audio means in practice

Null audio means the row is not shipping a native voice track you can rely on. You may get a silent master or placeholder silence. Either way, dialogue is your job downstream. That is fine when you want control. It is painful when someone assumed "video model" implied talking talent.

Open the catalog row before you brief talent style. wan-2-7-text-to-video should be chosen because you want that picture path, not because you hoped audio would appear. If you need native audio in-family, check sibling rows instead of wishing.

Silent plates are excellent for caption-led ads, product motion, and B-roll. They are incomplete for lipsync testimonials until you add voice. Name the deliverable correctly in the ticket: "silent plate" versus "speech-ready UGC."

Catalog notes beat forum memory. Someone will swear Wan always talks because they used a different version last quarter. Believe the row you are about to pay credits for.

Building the speech pipeline on purpose

Picture lock the silent plate first. Write or finalize the script second. Generate TTS third. Lipsync fourth. Captions after lipsync, because mouth timing can shift words. That order is spelled out in captions after lipsync, not before for a reason. Captioning before lipsync creates double work.

Pick a TTS voice that matches the face you generated. Mismatch is uncanny even when each part is good. Keep a small brand roster of voices so weekly ads feel related. Style-lock when you can.

Use AI lipsync as the dedicated step rather than asking the video model to invent speech it did not bill for. Separate tools keep failure modes diagnosable: bad plate, bad voice, or bad sync.

Budget credits for all steps in the estimate. A "cheap silent row" becomes a normal UGC cost once TTS and lipsync join. Honesty up front prevents fake savings narratives.

Where teams get blindsided

Sales demos sometimes play a talking sample from a different model family, then production picks Wan 2.7 for price. The client remembers talking. The row is silent. Align demo row and production row.

Batch pipelines fail when silence is not a first-class field in metadata. Tag audio=null from catalog into your DAM automation so schedulers cannot treat the plate as finished dialogue content.

Editors under time pressure add music and ship. Music is not speech. If the hook is verbal, music-only is still a miss on mute unless captions carry the claim. Captions can rescue a silent plate. They cannot invent a voice the brand promised.

A weekly workflow that respects silence

Monday: choose rows with audio fields visible. Tuesday: generate silent plates for tests. Wednesday: select winners, run TTS and lipsync only on winners. Thursday: caption and safe-area check. Friday: schedule burned masters. This cadence stops you from lipsyncing losers.

Keep a short checklist on the Wan 2.7 brief template: null audio acknowledged, TTS owner, lipsync owner, caption owner. Four checkboxes prevent Friday archaeology.

If speech is mandatory and you refuse multi-step pipelines this week, pick a row with native audio instead. Tools should match appetite. Forcing silence into a talking promise makes everyone cranky.

When you want catalog-honest Wan plates plus later speech tools in one studio habit, create a Versely account and plan the pipeline before the first render.

Producers should also sample rate and loudness plan before lipsync. A gorgeous plate with mismatched VO levels still feels amateur. Normalize voice, duck beds, then sync. Audio craft remains human even when the plate was generated.

If you A/B native-audio models against Wan 2.7 silent-plus-TTS, label the pipelines in analytics. Otherwise you will attribute wins to the wrong family name. Pipeline tags are part of creative learning.

Agency SOWs should list steps. 'AI video' as a single line item hides silence. Itemize plate, voice, sync, captions. Clients can cut scope consciously instead of being surprised by missing speech.

Write the decision in the brief so the next teammate does not reopen a settled choice. Short notes prevent repeat spend. Calm teams leave trails. Anxious teams regenerate the same lesson. Keep the trail next to the slug and the credit estimate so production stays boring in the best way.

FAQ

Do Wan 2.7 T2V and I2V rows include native dialogue?

Catalog rows are noted with null audio. Plan TTS and lipsync if speech is required. Always re-check the row before a major buy.

Can burned captions replace TTS for silent Wan plates?

Captions can carry the claim on mute without TTS. They do not create a voiceover. Choose based on whether the brand needs heard speech.

When should I lipsync versus leave the plate silent?

Lipsync when mouth-visible talking is part of the sell. Leave silent for B-roll, hands-only demos, or caption-led hooks where a voice would not add enough to justify the steps.