What to Ask For From an Audio Post House.

The deliverables list

Put these in the order or the SOW. Anything not on the list is a change request later, at whatever the day rate is then.

• Final mix, at the loudness target for the platform and territory, one file per version
• Separated stems — dialogue, music, effects, and voiceover kept apart rather than bounced together
• M&E stem (music and effects, no dialogue) if the spot will ever be dubbed for another territory
• Every duration you will traffic — 60, 30, 15, 6 — each mixed and measured on its own, not trimmed from the master
• Aspect-ratio and platform variants — vertical and square social cuts mixed independently, because a mix balanced for a television is wrong on a phone speaker
• File specification in writing, normally 24-bit 48 kHz WAV for broadcast and video
• The project archive, or a clear statement that you are not getting one

Why stems are the item to insist on

Stems are the single deliverable people skip and the single deliverable that decides what a late change costs. A stem set is the mix kept in separated groups rather than flattened into one file: dialogue on its own, music on its own, effects on their own.

With stems, the client who comes back in four months wanting the same spot with a different voiceover, or the music pulled under a new legal line, or a regional dub, gets an afternoon of work. Without them, the only way to make that change is to rebuild the mix, and if the studio has archived or the session has aged past its plugin versions, sometimes it is a re-record.

The M&E stem is the same argument for international versioning. A spot delivered with music and effects separated from dialogue can be voiced in another language by anyone. A spot delivered as a flat master cannot be, at any price, without redoing the mix.

Versions are where budgets actually go

The mix is rarely what overruns. Versioning is. A campaign that was scoped as one 30-second spot routinely ships as a 30, two 15s, a 6-second bumper, a vertical cut, a square cut, a radio edit and two regional variants — eleven deliverables from one line in the budget.

Each of those is real work, not an export. As covered in the loudness reference, integrated loudness is duration-dependent, so a 15 lifted from a compliant 30 usually does not measure the same and has to be re-measured and often re-balanced. A vertical cut is not the horizontal mix in a different container; the picture has changed, the emphasis has changed, and frequently the music edit has to change with it.

The fix is not to negotiate harder. It is to count the versions honestly at brief stage and price them, because they are going to exist either way.

The four questions to ask before you sign

None of these is adversarial. Each one is cheaper to ask now than to discover in week three.

1. 01What loudness target are you delivering to, and who confirmed it?

   US broadcast, European broadcast, streaming and theatrical are four different numbers. A studio that cannot answer this without checking has not read your distribution plan.

2. 02How many revision rounds are included, and what counts as one?

   The ambiguity is never the count, it is the definition. A round of consolidated notes from one stakeholder and four separate rounds from four stakeholders are the same work to you and very different work to the studio. Define it as consolidated notes and both sides are protected.

3. 03What happens when picture changes after the mix is locked?

   It will. Ask whether a re-conform is inside the fee or billed, and at what granularity — a two-frame trim and a re-edit are not the same event.

   • Agree a threshold in writing rather than a principle
   • Ask how long a re-conform takes, not only what it costs

4. 04Who owns the master, the stems and the session?

   These are three separate questions and the answers are often different. Owning the master while the studio retains the session is common and usually fine — but you need to know, because it decides whether you can take the work to another studio later.