Audio Engineering and Sound Design

01Purpose and scope

• This manual describes how audio engineering and sound design work is actually carried out on productions such as films, games, podcasts, commercials, interactive installations and live events. It covers the path from capturing sound or designing it from scratch through editing, mixing, and delivery to a client, broadcaster, platform or playback system. It is written as a reference for working engineers, sound designers, assistants and students who need a shared picture of the tasks, decisions and outputs involved.
• The scope is practical rather than theoretical. It does not teach acoustics, psychoacoustics or music theory in depth, though those underpin much of the work. It also does not set legal or regulatory requirements for any market. Where a rule depends on jurisdiction, such as loudness limits for broadcast or hearing exposure limits in a workplace, this manual points to the authoritative source and expects the reader to confirm current requirements.
• Audio work varies widely by context. A dialogue editor on a scripted series, a foley artist building a creature vocal for a game, and a live sound engineer mixing a conference keynote share some habits but face different constraints, schedules and failure costs. This manual notes those differences where they change the method.

02Core principles

• The first principle is that every sound has a job. A sound effect, a music cue, a room tone bed or a voice treatment should support a story, an interface, a message or a physical environment. Engineers who keep asking what the listener needs to hear usually make better choices about processing, level and placement than those who work from habit or plugin presets.
• The second principle is that the signal chain is only as good as its weakest documented link. Gain staging, sample rate, bit depth, file naming, time code, and the version of a session all affect the final result. Keeping these settings consistent and recorded prevents problems that are very hard to diagnose later, such as a mismatch in frame rate between picture and audio.
• The third principle is that listening conditions decide outcomes. A mix that sounds balanced in a treated room may fall apart on a laptop speaker, in a car or through a phone. Reference playback on several systems, and at more than one listening level, is standard practice. A mix that only works on the monitors that produced it is not finished.
• The fourth principle is that the deliverable defines the work. Loudness, channel format, stem structure, dialogue intelligibility targets and metadata are set by the receiver. Engineers should learn those requirements before they begin, not discover them during final delivery.

03Planning and preparation

• Start by reading the brief, script, picture lock or design document and listing every audio element the project needs. For a film, this includes production dialogue, dialogue replacement, sound effects, backgrounds, foley, music, and the final mix stems. For an interactive project, it also includes implementation logic, such as how many variations a footstep needs, how sounds are triggered, and what happens when the player is in a loud or quiet area. Write this list as a shared document so producers, editors and programmers can see what is in scope.
• Confirm the technical specification before the first session is opened. Ask for sample rate, bit depth, frame rate if the work is tied to picture, channel layout, file format, loudness target and true peak limit, stem breakdown, and naming convention. Where a broadcaster or streaming platform publishes delivery requirements, obtain the current version directly from them. Loudness targets such as the EBU R 128 recommendation are widely used in Europe and elsewhere, but the exact target for a given deliverable is set by the receiver.
• Plan the schedule around review rounds and the time needed for recording or design. Field recording may depend on weather, locations and talent availability. Sound design for a game may depend on animation or code being ready. Budget time for technical checks and for at least one full pass on the final playback chain. Experienced engineers often reserve more time for the last mix than the first, because late changes concentrate there.
• Set up the session template before work begins. Define track naming, bus structure, routing, color coding, reference tracks and metadata fields. A consistent template lets a second engineer pick up the work without guessing what each track contains.

04Recommended workflow

1. 1
   : Capture or gather source material. For recorded sound, select microphones, placements and preamp settings to suit the source and room, and log each take with scene, take number, mic and notes on noise or distraction. For designed sound, collect reference recordings, note the sonic qualities the director or designer wants, and record the source files with their origin and license terms.
2. 2
   : Ingest and organize. Import files into the session or library with consistent names, check that sample rates and bit depths match the project, and confirm that time stamps or timecode align with picture where needed. Back up the media to at least one separate location before editing begins.
3. 3
   : Edit and clean. Remove unwanted noise, clicks and breaths where appropriate, tighten timing, fix edits in dialogue, and build sound effect layers. Use spectral repair, noise reduction and similar tools with restraint, since over processing creates artifacts that are obvious on good playback systems. Keep unprocessed versions available so decisions can be reversed.
4. 4
   : Design and build. For sound design, layer source elements, shape envelopes, apply pitch and time manipulation, and place sounds in space using panning, reverb and distance cues. Test each element in context with picture or game footage, not only in isolation, because a sound that works alone may mask important information when everything plays together.
5. 5
   : Balance and mix. Set levels for each group, such as dialogue, music, effects, and ambience, against the reference. Address masking between elements, control dynamics where the medium requires it, and check mono compatibility. Mix at a level close to the one the audience will hear, and review the mix at low volume to confirm that dialogue remains clear.
6. 6
   : Master or prepare stems and deliverables. Apply final loudness and true peak control to the target format, create the required stems and versions, such as a full mix, a music and effects mix, and a dialogue stem for foreign language work, and render files with the specified codec and channel layout. Keep the master chain simple and documented.
7. 7
   : Verify and deliver. Play every deliverable in full on the reference systems, check file integrity, confirm names and metadata against the spec, and send a delivery note listing what was included, the loudness measurement, known issues, and the session version. Archive the session and the rendered files together so a later revision can be produced without reconstruction.

05Quality checks, risk controls and compliance

• Quality checks should be built into each stage rather than saved for the end. At the edit stage, check for clicks, phase problems, clipping and unintended silence. At the mix stage, measure loudness with an integrated meter, check true peak values, confirm the phase correlation of stereo material, and listen for the balance in mono. Typical indicators include integrated loudness against target, true peak margin, dialogue intelligibility in a reduced playback test, and the number of files that fail automated validation.
• Hearing safety is a real risk in this work. Long sessions at high monitoring levels, headphone use with loud material, and exposure during live sound or field recording can all contribute to hearing loss. Set monitoring levels with a reference such as a calibrated meter, take regular breaks, and use hearing protection on location and in live rooms. Workplace noise exposure limits depend on jurisdiction. In the United States, for example, the Occupational Safety and Health Administration publishes a noise standard, and employers and engineers should consult the current text and their local requirements.
• Intellectual property and licensing require attention. Sound libraries, sampled material, licensed music and recordings of people all carry terms. Keep records of the license, the permitted uses, and any restrictions on territory or duration. Recordings of identifiable people usually require consent and a release, and the rules differ by location and by the kind of use. Where rights are unclear, escalate to the production's legal or rights contact rather than making a judgment alone.
• Accessibility is part of quality. Dialogue clarity, appropriate loudness for broadcast, and sensible use of high frequency content for listeners with hearing differences all affect who can follow the material. Where a platform or regulation sets accessibility requirements, follow the current published guidance.
• This manual does not replace qualified professional judgment, approved procedures or required authorization. Where work touches medical audio, safety alarms, emergency broadcast, or regulated product sounds, follow the governing standards and the people responsible for those systems.

06Collaboration and documentation

• Audio work depends on many other roles. Directors and producers set creative priorities, picture editors control timing, game programmers control triggers and mixing systems, and clients approve or reject final material. Engineers should identify a single contact for approvals and for technical questions, and confirm which version of the picture, script or build the audio is synchronized to.
• Documentation is what makes a project recoverable. Keep a session log that records dates of changes, the reason for each major decision, the version sent for review, and the feedback received. Name files with a version number and a short description. Store reference mixes and approved versions in a location that the whole team can access. Notes on plug-in settings and processing chains are useful when a later revision needs to match the original sound.
• Feedback should be translated into specific actions. A note that says the effects feel weak is hard to act on. A note that asks for the door slam to sit 3 decibels louder against the dialogue, or for the creature vocal to lose its low end below 150 hertz, gives the engineer something to test. Where the client's language is subjective, ask for a reference example from another project, and confirm the agreed interpretation in writing.
• Handoffs need the same care. When a mix passes to another engineer for a language version or a platform-specific master, provide the session, the stem list, the target specifications and a note on what must not change.

07Measures and continuous improvement

• Measure the work in ways that reflect outcomes. Useful indicators include delivery pass rate on the first submission, number of review rounds before approval, loudness and true peak compliance across deliverables, the number of technical defects found after delivery, and the time spent on rework caused by missing specifications. For game audio, track memory use, voice count and the frequency of audio bugs found in testing. These numbers vary by project, so compare them against a team's own history rather than against outside figures.
• Review completed projects in a short debrief. Ask which decisions took the most time, which specifications were unclear, which tools produced unexpected results, and where the feedback loop broke down. Record the findings in a shared template and update the session template, the spec checklist and the delivery procedure where the debrief shows a repeated problem.
• Keep up with tools and standards without chasing every new plug-in. Test a new tool on a non-critical project first, and note its effect on the signal chain and its behavior under deadline pressure. Where a standard is revised, check the current version and update the procedure, rather than relying on memory of an older rule.

08Common mistakes and how to prevent them

• One common mistake is mixing too loud because the room feels exciting. The fix is to set a reference level, check the mix at a lower level, and compare it against a known track. Another is over processing dialogue, which often shows up as harsh sibilance, hollow tone or obvious breathing artifacts. Make small moves, bypass processing often, and compare the processed and unprocessed versions.
• Failing to check the final render is a frequent cause of delivery failure. Files can drop channels, use the wrong sample rate, carry the wrong stem name or exceed a true peak limit even when the mix sounded right in the session. Play the rendered files, not the session, and run an automated check where available.
• Working without confirmed specifications leads to rework. Engineers sometimes begin editing with a rough idea of the format and then discover that the broadcaster requires different loudness or stem structure. Confirm the spec in writing before the work starts.
• Poor file organization can make a project unrecoverable. Keep naming conventions, backups and version records in place from day one. Avoid storing the only copy of a session on a single drive or in a temporary folder.
• Ignoring the listener's environment is another risk. Test on small speakers, earbuds, a television or a car where relevant, because those systems shift the balance in predictable ways.

09Practical checklist

• Confirm the audio deliverables, formats, loudness target, true peak limit and stem structure in writing before editing begins.
• Set up the session template with naming, routing, reference tracks and metadata fields.
• Log every recording take or design source with its origin, license terms and relevant notes.
• Back up raw media to a second location before any editing or processing.
• Check sample rate, bit depth and timecode alignment against the project specification.
• Check mono compatibility and stereo phase at the mix stage.
• Measure integrated loudness and true peak on every master and stem.
• Play each rendered deliverable in full on at least two playback systems, including one that reflects the audience's likely environment.
• Confirm rights, releases and consents for all licensed material and recordings of people.
• Verify hearing safety measures, monitoring levels and breaks during long sessions.
• Send a delivery note with file list, measurements, known issues and session version.
• Archive the session, reference mixes and final files together with version notes.

★Learning sources

• Micro Degree: Audio Engineering and Sound Design
• Micro Degree: 2D Animation Fundamentals
• Micro Degree: 3D Animation Techniques
• Micro Degree: Illustration Fundamentals