A blank script stand,pop filter and abstract sound ribbons illustrating voice-over preparation.
Original AI-generated editorial illustration; not a real account, product sale or completed job. View full-size image.

A voice-over offer should define the script, listener, performance direction, pronunciation decisions and delivered audio. Agree on those points before recording the complete job. Owning a microphone or producing a fluent read does not establish that the performance fits a client's use or that an audition will lead to paid work.

This guide proposes a preparation workflow for an original spoken recording. Fovza has not performed a customer's script, qualified for a marketplace or verified paid bookings. The practice exercise is invented. It is distinct from podcast editing, which starts with supplied recordings and an edit brief.

Establish the script and its intended use

Ask who the listener is and what the recording needs to communicate. A welcoming instruction, a training explanation and a character performance call for different direction. Identify the words that must remain exact and any parts that are still awaiting approval.

Record the source version and the person authorized to approve it. If the wording changes during recording, identify the revised passages and the effect on scope. Do not quietly join takes from different script versions and assume that the resulting message is still correct.

Clarify the intended uses and who may receive the files. Do not treat permission to record one script as blanket permission for unrelated reuse, synthetic-voice training or imitation of another performer. Resolve those requests separately before supplying material; this article is not a rights agreement or legal template.

Resolve pronunciation and direction before the session

Voices' script guidance recommends clear direction about tone, role and audience, and suggests pronunciation recordings or phonetic notes for unfamiliar words. The article dates from 2020. We use it for general preparation questions, not as a current booking, rate or marketplace-admission rule.

Source: Voices' guide to preparing voice-over scripts. Read 2026-10-26. No provider story, performer identity or example script is adopted here.

Create a query list for names, abbreviations and phrases that could be read in more than one way. Ask the client to confirm the intended pronunciation and meaning. An internet search may not resolve a new product name or an internal term.

Agree on a short reference read when appropriate. Ask whether the pace, emphasis and level of energy fit the listener. Save the decision with the relevant script version. A reference helps establish direction; it does not guarantee that every later line will need the same delivery.

Rehearse meaning as well as timing

Read the script aloud before planning the full session. Notice where the thought changes, where a listener needs a pause and where the wording is difficult to say naturally. Ask for clarification if a sentence is ambiguous rather than solving it with an emphasis that changes its meaning.

If a fixed duration is required, test a representative read at a comfortable pace. Do not assume that dividing a word count by a generic speaking rate establishes the actual duration of the finished performance. Allow for pauses and the intended delivery, then resolve a mismatch with the client.

Keep script editing separate from performance unless both are included. Suggesting that a sentence is hard to speak can be useful feedback, but the client may need to approve any wording change. Preserve a clear record of what was authorized.

Define revisions before quoting

Voices' revision-policy page describes changes to scripts, performance feedback and additional takes as possible revisions. It allows performers to describe their policy in a proposal and does not prescribe one universal revision charge. Its sample policy is an example, not a rule that automatically applies to an independent agreement.

Source: Voices' revision-policy explanation. Read 2026-10-26. This guide has not submitted a proposal or tested a provider payment process.

Write the included review in plain terms. Distinguish a missed word or agreed pronunciation correction from new copy, a changed audience or a different creative direction. State how feedback should be consolidated and when a new script version becomes a new task.

Count preparation, pronunciation questions, recording, listening, file preparation and communication in the proposed workload. A minute of finished audio is not necessarily a minute of labour. This guide does not supply a market rate, booking forecast or expected earnings.

Prepare a controlled practice recording

Write your own short welcome message for an invented community workshop. Include one term that needs a pronunciation decision and a clear change of thought. Keep the wording outside a real brand, copyrighted script or specialist claim you cannot substantiate.

Record a brief test and listen before continuing. Check whether the environment and recording setup let the listener understand the words without distraction. If you cannot produce an acceptable sample in that space, address the limitation before offering the same setup for a client job.

For the exercise, make two intentionally different reads and explain which direction each follows. Keep the script version and take identifiers with the files. The point is to demonstrate a reviewable performance choice, not to claim that Fovza completed a commercial voice-over.

Inspect the audio you will deliver

Listen to the complete chosen take while following the approved script. Check missing or repeated words, names, unwanted interruptions and any joined passages. A technically clean recording can still be the wrong performance or contain the wrong wording.

Open the actual delivery files and check their names, order and agreed format. If the client requires separate lines or sections, confirm that each file corresponds to the correct text. Keep any unapproved alternate takes out of the final package unless the brief includes them.

Ask for the client's current technical requirements rather than assuming a universal audio specification. If editing, processing or file preparation exceeds your competence, define that limit. A successful save or export message is not proof of acceptance in the client's production workflow.

Separate the export settings before you quote

Ask the client to name the required sample rate, bit depth and channel layout. Sample rate describes how often the sound is sampled; bit depth describes the precision of each sample. They are different settings. Audacity's digital audio explanation also distinguishes lossy compression, which discards audio information, from lossless compression. An MP3 bitrate is not the same setting as the bit depth of uncompressed audio.

For the invented workshop exercise, write a delivery note with separate fields for file format, sample rate, bit depth where applicable, mono or stereo, and the approved take. Ask the reviewer to complete the required values before recording. Treat an unanswered field as a question to resolve, rather than copying settings from an unrelated job.

Keep the original recording while making the delivery copy. Converting an MP3 into a lossless file does not bring back information already discarded by lossy compression. Reopen the exported copy, inspect its properties and listen through it; checking the editor's project settings alone leaves the delivered file unchecked. If the client changes the specification, return to the retained original and reassess the export instead of treating a new file extension as proof that the requirement is met.

Keep approval separate from publication

Deliver the agreed files with the script version and a short note identifying outstanding questions. The client should be able to review what was recorded without guessing which take is current. Keep release, account access and distribution outside the assignment unless separately authorized.

Agree on secure file handling and retention. Do not upload a private script or recording to another service for assistance without permission, and do not reuse it as a demo merely because the project is finished. For other work models, see remote side hustles. Your next step is an original, checked sample with clear direction and revision boundaries, not a promise of paid auditions.