AI Film Sound: Voice, Music and Mix From Absolute Silence

An AI film arrives with no production sound at all, so the soundtrack gets built from silence in seven passes: spot the cut, lay continuous ambience beds, place dialogue, add Foley, license music, mix to dialogue, deliver stems and a cue sheet. Ambience comes first, because shots generated separately never shared a room.

No room tone, no footsteps, no traffic through a window, nothing recorded on the day because there was no day. A live-action mixer inherits a continuous recording of a real space and spends post cleaning it up. You inherit forty clips that each exist in their own acoustic vacuum, and what gives an AI film away is rarely the quality of the elements. It is the fact that they were never in the same room.

This sits under the complete AI filmmaking guide as the audio half of post. The seven-day production log gives sound a full day and this is what goes in it.

Native model audio is a scratch track, not a mix

Veo 3 and the models that followed it generate synchronised audio in the same pass as the picture: dialogue, effects, ambience, together. It is genuinely useful, and you still cannot mix with it.

What comes back is one mixed audio track inside the MP4. There are no dialogue, music and effects stems, so there is no way to lower the music without also lowering the footsteps, no way to keep the ambience while replacing a line, no way to duck anything under anything else. You own a bounce somebody else made, at a balance you did not choose.

Worse for continuity: each clip's generated ambience is its own invention. Two shots in the same kitchen come back with two different kitchens behind them. Cut them together and the background jumps on every edit. Viewers do not identify this consciously. They just feel that the film is a sequence of clips rather than a scene.

So use it for three things and nothing else:

• A timing reference while you cut. The generated audio is sample-accurate to the picture, which makes it a free scratch track.
• A sound-effects sketch. If the model's version of a door suggests the texture you want, go find that texture.
• Lip-sync anchoring, where you are generating to a line rather than dubbing onto a finished shot. The AI voices and lip sync article covers that case properly.

Then mute it. Every one of them. Before you place a single intentional element, the timeline should be silent.

The pass order, and why ambience comes first

Most sound tutorials start with dialogue, and for a live-action film that is right: the dialogue is already there, recorded, and everything else is built around it. For generated footage the order flips at the front.

Build the ambience bed first. One continuous bed per location, laid across every shot in that location including the cuts, running underneath the whole scene rather than per clip. That single continuous layer is what tells the ear the shots share a space, and it does more for the illusion than any other element you will add. It is also the one thing the model cannot give you, because the model works a clip at a time and a bed is by definition longer than a clip.

Dialogue still comes first for the cut. Place lines, trim picture to the performance, lock the edit. The distinction matters: voices drive your editorial decisions, ambience drives your mix structure. Lock picture against the voices, then build the mix from the bed upward.

Pass	What you build	Input you need	Output
1	Spotting list	Picture-locked cut	Timecoded list of every sound the film needs
2	Ambience beds	One bed per location, min. 2 min each	每个拍摄地点铺设一条连续的环境底噪，最短 2 分钟，作为每个镜头下方的持续声音层，即使画面剪切也保持连贯
3	Dialogue	Generated or recorded lines, per character	Lines placed, cut trimmed to performance, picture locked
4	Hard effects and Foley	Per-action library or recordings	画面中每一个会发出声响的动作都需要对应的音效素材，可来自音效库或现场录音
5	Music	Cleared tracks, theme plus alternate	已取得授权的音乐曲目，包含主题曲及备选方案，按 spotting list 中的标注点放置
6	Mix	All of the above on separate tracks	Balanced mix, dialogue intelligible, levels to spec
7	Deliverables and paperwork（交付物与文档）（交付物与文档）	Final mix（最终混音）（最终混音）	Master 主文件、stems 分轨素材、cue sheet 提示清单

Seven passes, and for a six-minute film they are roughly a day's work if you have your elements ready and three days if you do not. The spotting list in pass 1 is what decides which of those it is.

Spotting: watch it silent, write down every noise

Play the locked cut with all audio muted and write down, with a timecode, every single thing that should make a sound. Work from what the picture implies rather than from the library you already have.

A character sets a glass down: that is a cue. A car passes behind a window in the background of shot 14: cue. A door is already closing when the shot starts, meaning the latch click lands four frames in: cue. You will write thirty to sixty lines for a six-minute film, and the list will be longer than you expected, because generated footage is full of incidental motion nobody planned.

That list becomes your build order, your budget reference, and your diagnostic tool for catching sound problems that originate in the picture itself. When shot 22 shows a character's mouth moving in the background of a two-shot but you have no corresponding line scripted for them, you've identified that continuity gap now—during spotting—instead of discovering it later in the final mix when fixes are exponentially more expensive.

Preserve the shot numbering system you established during the edit so that the spotting list, the take log generated during your dailies review, and the final cue sheet all reference identical shot identifiers. Renumbering shots as they pass between departments—editorial to sound to mix—is the single most common workflow break that causes a mix session to lose half a day hunting for sync points and reconciling mismatched references.

Foley for motion nobody performed

Hard effects are the ones tied to a visible, concrete source onscreen: the door closing, the gunshot report, the phone ringing. Foley, by contrast, is the quiet textural layer underneath all of that—the sound of a body moving through a space, interacting with surfaces and fabrics. Generated footage, because it lacks the incidental acoustic information captured by a production microphone, requires far more deliberate Foley coverage than live-action material does.

A camera on a real set records the actor's clothing, their weight on the floor, their hand on the table. All of it arrives free, buried in the production track, and the audience never notices it until you take it away. Generated footage has none of that, so a character can walk across a room in total silence while looking entirely convincing. The picture is fine. The scene feels dead.

Footsteps and cloth movement are the two Foley categories that matter most for perceived realism. Get those elements properly synced under every walking shot and the figures immediately gain physical weight and spatial presence. Once that foundation is in place, then handle whatever additional props, touches, and movements your spotting list caught during the initial pass.

A practical constraint specific to generated content: generated motion is often slightly inconsistent in timing from frame to frame, with a step that lands a frame early or a hand that arrives sooner than the physics would suggest. Because of this variability, cut each effect to the picture frame by frame rather than laying down a rhythmic footstep loop and hoping it stays in sync. Loops reveal the drift and call attention to the mismatch. Individual placement, adjusted per footfall, hides it and maintains the illusion of weight and timing.

Music, and the licence you can actually prove

This is precisely where AI-generated films encounter their most significant licensing challenges, and the root issue is almost never about the musical quality itself being substandard.

The most straightforward option for a generated film production is to utilize an AI music generator, yet the licence terms attached to the subscription plan you are likely to purchase may not actually cover the specific type of film project you are producing. ElevenLabs structures its music rights across multiple tiers, and this licensing ladder progresses from personal use through two distinctly different grades of commercial authorization.

The commercial use row of the Music pricing table on elevenlabs.io/music, captured 1 October 2026.

"Commercial, not enterprise" represents the critical row that filmmakers need to examine closely, and the Eleven Music Model-Specific Terms explicitly detail what categories of use remain excluded. Every self-serve subscription plan spanning from Free through Business, plus Enterprise Music Lite, authorizes "All online and offline commercial use permitted, except film, TV, radio, & Studio Games". Only the full Enterprise tier removes that specific carve-out restriction entirely. The terms page carries a date of 26 May 2026, so it is essential to verify it again before you depend on this information for your production decisions.

Traditional stock music libraries establish their licensing boundaries at this same demarcation point. Epidemic Sound places online ads, branded content and social media client work within its Pro plan offerings, and its dedicated filmmaking page explicitly states that "films intended for theatrical/festival release" require the Enterprise plan's customized tailored licence arrangement.

So both of the fastest music routes license you for the internet, and the thing most AI filmmakers want next is a festival. If you are working through the AI festival submission list, the entry form will ask you to warrant that you hold the rights to every element in the film—meaning you personally possess the legal clearance or license grant for every visual asset, dialogue sample, sound effect, and musical track—and a social-tier music subscription does not give you that warranty because those plans typically restrict use to personal, non-commercial streaming and exclude public exhibition or theatrical screening.

Three ways out, in the order most people should consider them based on budget, timeline, and the scope of rights you actually need for festival submission and potential distribution:

Route	Covers festivals	Cost shape	Catch
Composer, work for hire	Yes, if the contract assigns or licenses it to you and explicitly grants festival exhibition and distribution rights and explicitly grants festival exhibition and distribution rights	Per project, negotiated based on scope, usage term, and exclusivity based on scope, usage term, and exclusivity	Needs lead time and a written agreement that specifies deliverables, payment schedule, and clearance for the territories and formats you intend to enter that specifies deliverables, payment schedule, and clearance for the territories and formats you intend to enter
Library or AI tool, upgraded plan	Yes on the tier that says so—look for language like "broadcast," "film festival," "theatrical," or "all media" in the license summary—look for language like "broadcast," "film festival," "theatrical," or "all media" in the license summary	Higher subscription tier or custom quote, often labeled Pro, Enterprise, or Film & Broadcast	Verify the tier before you cut to the track, download the license certificate, and keep a copy with your festival paperwork in case a distributor or rights buyer asks for proof, download the license certificate, and keep a copy with your festival paperwork in case a distributor or rights buyer asks for proof
Public domain or permissive Creative Commons	Depends entirely on the specific licence—CC BY and CC0 are generally safe for festivals, but CC BY-NC (NonCommercial) may disqualify you if the festival charges admission or offers prize money, and CC BY-SA (ShareAlike) can create downstream licensing headaches for distributors—CC BY and CC0 are generally safe for festivals, but CC BY-NC (NonCommercial) may disqualify you if the festival charges admission or offers prize money, and CC BY-SA (ShareAlike) can create downstream licensing headaches for distributors	Free to low cost, though you may pay a nominal fee for a high-quality recording or master file cost, though you may pay a nominal fee for a high-quality recording or master file	Attribution terms vary by license, and remember that sound recordings are copyrighted separately from compositions, so a public-domain Chopin piece performed and recorded in 2023 is still under copyright for the recording itself

That last catch traps people constantly. A Debussy composition is public domain. A modern orchestra's recording of it is not, and you need clearance on the recording as well as the composition. Check both. The composition copyright and the recording copyright are two separate legal entities—even when the underlying work entered the public domain a century ago, a 2020 recording by the London Symphony Orchestra remains protected, and sync licensing requires you to clear rights with both the publisher (for the composition) and the label or performer (for the master recording). The composition copyright and the recording copyright are two separate legal entities—even when the underlying work entered the public domain a century ago, a 2020 recording by the London Symphony Orchestra remains protected, and sync licensing requires you to clear rights with both the publisher (for the composition) and the label or performer (for the master recording).

One theme and one alternate is enough for a short. Twenty library tracks is what a film sounds like when nobody decided what it was about. A cohesive score built around a single melodic or harmonic identity gives the audience an emotional anchor; when you cycle through dozens of unrelated cues, the soundtrack becomes a slideshow that fragments the narrative instead of supporting it, and the film loses its tonal center. A cohesive score built around a single melodic or harmonic identity gives the audience an emotional anchor; when you cycle through dozens of unrelated cues, the soundtrack becomes a slideshow that fragments the narrative instead of supporting it, and the film loses its tonal center.

Try it free

Try 熙熙 for free on your script

It is free. Import your existing project, get a clearer view of your outline, and regain control of your story structure in minutes.

Start Free

The mix: get dialogue right, then everything else

Separate tracks per element, one fader group each for dialogue, music and effects. Mix dialogue first, to intelligibility, then set music and effects against it. Reverse that order and you get a beautiful score over a film nobody can follow. Dialogue intelligibility is the foundation of the mix hierarchy—audiences will tolerate rough effects or thin music, but unintelligible speech breaks immersion instantly, so establish clear, present dialogue levels before layering in the score and the Foley, and use your music and effects faders to support rather than compete with the spoken word. Dialogue intelligibility is the foundation of the mix hierarchy—audiences will tolerate rough effects or thin music, but unintelligible speech breaks immersion instantly, so establish clear, present dialogue levels before layering in the score and the Foley, and use your music and effects faders to support rather than compete with the spoken word.

Delivery numbers depend entirely on where the film goes, and the two destinations an AI short actually has ask for very different ones. Festival technical specs often call for theatrical-standard deliverables—23.976 fps ProRes 422 HQ, stereo or 5.1 mix stems, and sometimes a separate M&E track—while online platforms like YouTube or Vimeo accept H.264 at variable bitrates, integrated stereo audio, and far more relaxed loudness targets, so you need to know your exhibition path before you lock your export settings. Festival technical specs often call for theatrical-standard deliverables—23.976 fps ProRes 422 HQ, stereo or 5.1 mix stems, and sometimes a separate M&E track—while online platforms like YouTube or Vimeo accept H.264 at variable bitrates, integrated stereo audio, and far more relaxed loudness targets, so you need to know your exhibition path before you lock your export settings.

Destination	Target	Measurement
YouTube, Spotify, Amazon Music, Tidal	around -14 LUFS integrated, -1.0 dBTP ceiling	Integrated programme loudness, ITU-R BS.1770, measured across the entire programme duration to ensure consistent perceived volume and prevent clipping on consumer playback systems, measured across the entire programme duration to ensure consistent perceived volume and prevent clipping on consumer playback systems
Netflix delivery	-27 LKFS ±2 LU, peaks not above -2 dBTP	Dialogue-gated measurement, meaning the metering system actively ignores music layers and sound effects during analysis

Do not attempt to convert or reconcile the values between those two specification rows. They represent fundamentally different measurement methodologies rather than merely different target numbers: Netflix anchors its loudness standard specifically on dialogue content and applies dialogue intelligence gating during metering, which means music beds and sound effects layers never influence the measurement reading at all, whereas the streaming platform specifications normalise the entire programme mix as a whole, which in practice means a loud orchestral score or heavy sound design will consume your available dialogue headroom. Netflix additionally recommends that the complete final mix should sit within a 4 to 18 LRA range, with the dialogue elements themselves contained inside a tighter 10 LRA window.

If the film's final destination is a streaming platform delivery, creating a single universal master levelled around -14 LUFS integrated loudness with a -1.0 dBTP peak ceiling will cover the technical requirements of most platforms simultaneously. If a sales agent or distributor is handling the release, request their complete technical spec sheet before you begin the final mix session, not after the mix is already locked.

Then run three checks that catch nearly everything:

• Phone speaker. Play the whole film from a phone at half volume, no headphones. If a line disappears, the mix is wrong, whatever the meter says. This is how most of your audience will actually hear it.
• The cut check. Scrub each edit point with your eyes closed. Any audible jump in the background means your ambience bed is not continuous through that cut.
• Mute the music. Play a full scene with the score off. If it collapses, the music is carrying the scene and the sound design is not finished.

Where the mix usually goes wrong

In rough order of how often I see each one in AI shorts:

• Native model audio left running under the real mix. Two ambiences fighting, and nobody can hear why the scene sounds muddy. Mute everything the model gave you, deliberately, before you start.
• Ambience built per clip: forty short beds instead of three long ones, and a background that jumps on every cut.
• No Foley at all. Characters walk in silence through fully realised rooms.
• Music placed before picture lock. You cut to the track, the edit changes, and now the hits land nowhere. Music is pass 5 for a reason.
• Mixing on headphones only. Flattering, and it hides exactly the dialogue problems a phone speaker exposes.
• No stems in the deliverables. The festival wants a textless master or a different language version, and you have one baked file. Export dialogue, music and effects stems while the session is still open.

Deliverables and the paperwork nobody warns you about

Picture editorial departments complete their workflow by delivering a finished master file. Sound departments finish by delivering both a master file plus comprehensive technical and legal documentation, and it is specifically that documentation package that unblocks the distribution pipeline and clearance processes.

A music cue sheet lists every cue in the film with its timecode in and out, duration, usage type, composer, publisher and performing-rights organisation. Festivals ask for it, broadcasters require it, and it is how composers and publishers actually get paid when a film is screened or broadcast. It takes twenty minutes if you kept notes while placing cues and an afternoon if you did not. Build it from your spotting list with the cue sheet generator, which exports the PDF and CSV most submission forms accept.

One cue row in the 熙熙 cue sheet generator, captured 1 October 2026. Duration is computed from the IN and OUT timecodes.

The usage classification column in cue sheets is the metadata field that people consistently mishandle. Background instrumental cues, visual vocal performances and main title placements each carry different royalty calculation weightings in performing rights systems, so lazily marking every single music cue as background usage simply because it appears as the default dropdown option quietly costs your composer real licensing revenue every single time the film screens publicly or streams.

While you are there, finish the rest of the audio paperwork: stems exported, voice sources documented per character, music licences saved as files rather than as emails you will search for later, and generated audio disclosed where the platform asks. The AI film disclosure rules article covers what each platform asks you to declare, and synthetic voice is increasingly a separate question from synthetic picture on those forms—distributors and festival portals now often present voice generation and image generation as distinct checkbox fields, reflecting how different jurisdictions and platforms treat the two technologies under their submission guidelines.

Keep a one-page audio log next to the film: which voice model produced which character, which music came from where under which licence tier, which effects you recorded yourself. The first time someone asks you to prove a chain of rights, that page is the answer, and writing it after the fact is miserable. Include version numbers or generation dates for voice models, the exact plan tier or credit purchase that granted the licence, and the original source filenames for any field recordings or purchased sound effects libraries you layered into the final mix. Include version numbers or generation dates for voice models, the exact plan tier or credit purchase that granted the licence, and the original source filenames for any field recordings or purchased sound effects libraries you layered into the final mix.

FAQ

Can I just use the audio the video model generates?

As a scratch track, yes, and it is good at that. As a finished mix, no. You receive one mixed track with no separate dialogue, music or effects stems, so you cannot rebalance dialogue against score, adjust the loudness of a sound effect in isolation, or duck the ambience beneath a critical line of speech—and each clip's generated ambience is independently invented, which makes the background jump at every cut, creating jarring discontinuities in room tone, environmental noise profile, and spatial reverb character that betray the edit structure.

How long does sound take on a short AI film?

Budget one full day for a film under about seven minutes if your elements are ready, and three days if you are still sourcing music and recording Foley. Spotting is two hours, ambience beds are quick once you have the files, and the mix is where the time actually goes—level automation, EQ matching across cuts, reverb sends to unify disparate elements, and iterative playback to catch masking, phase issues, or moments where dialogue intelligibility drops all compound into the longest block of the post-audio schedule.

Do I need a composer, or is AI music fine?

Both work. What decides it is where the film goes. Check the licence tier against your distribution plan: on self-serve plans, AI music tools and stock libraries routinely clear online use while carving out film, TV, radio and theatrical or festival release—platforms such as Artlist, Epidemic Sound, and generated-music services often gate broadcast, theatrical exhibition, and physical media under higher-tier subscriptions or per-project buyouts. A work-for-hire composer with a written agreement is the cleanest route for anything headed to a festival, because you control the master recording and publishing outright, avoiding the mid-run discovery that your festival screening violates a platform's usage restrictions.

What loudness should I mix to?

If the film is going to YouTube or the music platforms, around -14 LUFS integrated with a -1.0 dBTP true-peak ceiling. If a distributor is involved, ask for their spec: Netflix, for instance, wants -27 LKFS ±2 LU measured dialogue-gated with peaks no higher than -2 dBTP, which is a very different mix.

How do I keep dialogue consistent when each shot was generated separately?

Treat it as one continuous recording rather than per-shot audio. Same voice settings per character across the whole film, dialogue sitting on one track group with one chain of processing, and a continuous ambience bed underneath so the lines share a space. The per-line craft, prosody and lip sync is covered in the AI voices article.

Do I have to declare AI-generated audio separately?

Often, yes. Several platforms and festivals now ask about synthetic voice as a separate question from generated picture, particularly where a real person's voice might be involved. Declare the voice tools by name alongside the video tools, and keep the per-character log so you can answer precisely.