How does audio production outsourcing work for a AAA game title?
It works as eight steps run against your milestones, with outside vendors supplying most of the assets. You scope the work and write an audio bible. Sound designers build the effects, a composer writes the score, and a voice vendor casts, directs and records the dialogue. Someone then wires it all into the game through Wwise, FMOD or the engine's own audio system, testers check it in the build, and the vendor hands over final mixes, stems and session files. Buying audio this way is growing: GameSoundCon's 2025 survey found people employed by audio outsourcing companies went from 5% of those working in game audio in 2016 to 14.7% in 2025.
What is the typical workflow when outsourcing game audio production?
The typical workflow follows the game's production phases. Li & Ortega, a game audio company, splits it into pre-production (audio style guides, temp tracks, planning), production (asset creation, voice recording, music composition) and post-production (mixing, mastering, bug fixing, optimization). It warns that implementation has to run throughout rather than at the end, because hooking sounds up properly takes time. Each milestone then repeats a short loop: you send a build and an asset list, the vendor delivers, you review in the game, and they revise. Agree the revision rounds up front, since Li & Ortega lists unclear contracts on revisions and ownership among the common pitfalls.
What goes in an audio bible?
An audio bible is the one document every audio vendor works from, and it should settle creative direction and technical limits before anyone makes an asset. Vendors use different names: Li & Ortega budgets "audio style guides" in pre-production, and Dynamedion lists "Audio & Music Book Creation" among its services. Put in the tone and references, temp tracks, naming conventions, the middleware and version, memory and voice-count budgets per platform, and loudness targets. Add the asset list with counts. If you have no audio lead to write it, buy it as the vendor's first paid milestone, because every later quote depends on it.
What should an outsourced audio vendor deliver?
Expect finished, named and mixed assets ready to implement, and ask for stems and session files as well. Demute says each asset is "cleaned, named, mixed and formatted" for integration into Unreal, Wwise, FMOD or your engine, and Echorbit Audio delivers scores as stems your team can move between game states. For voice, Altagram delivers files in your chosen format with a complete script that records text changes, so subtitles match. Implementation should leave you a middleware project your team can read: Flutu names clear events, optimized banks and documented parameters. Write the deliverable per asset into the contract, including revision rounds.
How much does it cost to outsource game audio?
Music is priced per finished minute, voice per session or line, and sound design per asset, per bundle or on a retainer. GameSoundCon's 2025 survey puts freelance composers at a median $1,700 per finished minute for AAA, $1,000 for mid-core and $407 for indie, so an hour of AAA-median music is 60 × $1,700 = $102,000. Li & Ortega publishes US voice costs that come to about $3,000 for a four-hour non-union session with studio and director, or $3,370 union, before editing. We found no published per-asset price for AAA sound design. Li & Ortega says audio usually takes 5% to 10% of a game's budget and advises 10% to 15% contingency.
How does adaptive music change the job?
Adaptive music changes the deliverable from finished tracks to a system of segments, layers and rules that the game plays back. Flutu describes music that shifts with gameplay through tested layers, states and transitions, and Audiokinetic describes Wwise cues that respond to rule-based transitions. So the composer has to write parts that layer and recombine cleanly, and someone has to author and test the transitions in the middleware. Li & Ortega says to budget for multiple versions, looping and alternate themes. Ask whether the per-minute rate covers stems and transitions, who builds the music system, and whether it gets tested in the running game.
Should a AAA title use Wwise or FMOD?
Wwise is the popular choice on AAA titles, though your game engine and budget often decide it. GameSoundCon's 2025 survey says Wwise stays popular on large-scale AAA work, games on Unreal Engine often use the engine's built-in audio and otherwise favour Wwise over FMOD, and FMOD was the most popular audio engine on Unity. Studios with their own engine tend to use Wwise or a custom system. Licensing differs too: above a $2 million budget Wwise starts at $25,000, and above $1.8 million FMOD costs $18,000 per game. Choose before implementation starts, because switching later means redoing the integration.
Who does the middleware integration, and how long does it take?
Either a technical sound designer at the vendor or your own audio programmer does it, so settle which before you sign. Many vendors sell it as its own service: Demute authors variations, RTPCs, interactive music systems, States and Switches in Wwise, FMOD or MetaSounds, then hooks them up to gameplay. Li & Ortega lists underestimating implementation time as a common pitfall. For scale, an older Audiokinetic customer article, on its blog since 2016, quotes Ensemble Studios taking three weeks to get Wwise in: about a week to run the engine, then a couple of weeks on parameters and basic events. Treat that as the setup only, because assets keep arriving for the rest of production.
Should SFX, music and voice come from one vendor or several?
One vendor suits a studio with little audio staff, and several suit a studio whose audio lead can run them. Li & Ortega describes the split: full-service teams handle leadership, music, SFX, voice and implementation, while specialists and freelancers cover single components such as orchestral recording or localization. One vendor saves handoffs, and Dynamedion pitches exactly that, with composers and sound designers adapting to each other in the mix. Several vendors let you pick a specific composer, or a voice house with the languages you need. Either way, keep the audio bible and the final mix decisions with one named person.
How much does voice add when a AAA game ships in several languages?
Voice scales with lines times languages, so it can match or pass the cost of translating the whole script. Transphere's guide says actor fees often start around $15 per line, and its example of 500 lines plus $500 of editing comes to $8,000 per language. It also says audio can easily double a localization budget. Many large games record fewer languages than they translate: Steam lists Baldur's Gate 3 with 15 languages and full audio in one, and Hi-Fi RUSH with 14 and full audio in nine. The game localization cost guide shows voice as its own line by word count and language count.
What does audio QA check on an outsourced project?
Audio QA checks the sound in the running build, on every target platform, against the limits in your audio bible. That covers missing or wrong sounds, the mix across gameplay situations, memory and voice counts, and loudness. Audiokinetic's mastering guide, written with PlayStation audio staff, gives PlayStation's recommendations as of February 2024. They are -24 LUFS integrated for console and -18 LUFS for mobile or portable, plus or minus 2 LU, measured over at least 30 minutes of varied gameplay. Flutu sells CPU, memory, voice count and bank profiling as its own line. For voiced languages, add subtitle sync checks, which Altagram and Testronic both list under localization QA.
Who owns the audio when a vendor makes it?
On AAA games the publisher or developer normally owns the music. In GameSoundCon's 2025 survey, none of the freelance AAA composers said their last game was done as a licence, against 11% for mid-core and 36% for indie, which leaves work for hire as the AAA norm. Li & Ortega notes that buyouts cost more than royalty-shared or usage-based licences, and that venture-funded and larger studios tend to prefer permanent buyouts. Get the transfer in writing, and ask the vendor to clear any licensed samples or libraries inside the deliverables. Make stems and session files part of the same transfer.
When should an audio vendor start on a AAA title?
Start in pre-production if the vendor is writing the audio bible or helping pick middleware, and by vertical slice at the latest for sound design. Dynamedion offers to consult on tools at an early development stage, and Li & Ortega warns against leaving implementation to the end. Music can start once tone and level structure are clear. Voice comes last, after the script locks, because pickup sessions pay for the same work again. Rush jobs can double the price, on Li & Ortega's figures, so a late start costs more as well as leaving less time to fix the mix.
