Two proposals may offer narration for the same approved script. One includes a person recording the performance. The other uses a text-to-speech system to generate it. A short demonstration can help assess the sound, but it does not reveal all the work, permissions and decisions behind the finished track.
Compare the production methods before choosing on apparent convenience. Ask who directs the result, who checks it, what use is authorised and what happens when the script changes. Neither method removes the need for an accountable production and review process.
Know which synthetic route is being proposed
Text-to-speech converts written input into synthesised speech. Microsoft's speech-service overview distinguishes its standard voices from custom voice options. That is one provider's product structure, not a rule that all synthetic voice services operate on the same terms.
Ask whether the proposal uses an existing authorised voice option or involves a voice derived from a particular person. The second raises a different permission and governance question. Do not infer that an earlier interview recording or ordinary voiceover booking authorises creation of a synthetic version of that person's voice.
This article concerns choosing an authorised narration workflow. It does not provide voice-cloning instructions, recommend imitating a named person or establish a right to use any provider's voice commercially. The actual voice source, account arrangement and intended use need checking by the responsible people.
Compare how the performance will be directed
With a recorded human voiceover, clarify the recording and direction arrangement. Who briefs the narrator, communicates the intended meaning and selects the approved performance? Ask how corrections or additional passages would be handled under the actual engagement.
With synthetic narration, clarify who turns the approved text into the proposed output, assesses the result and makes further passes where needed. What can be adjusted within the selected system, and what requires another approach? Do not assume that naming a desired tone guarantees it can be achieved for every passage.
Both methods need an intelligible, appropriate performance in the film. Avoid broad claims that human delivery is always superior or synthetic delivery is always sufficient. Ask the supplier to explain the process it would use for the specific script and audience.
Assign verification to a person
An approved script and an approved audio file are separate things. Someone needs to compare the resulting narration with the intended words and meaning, then check how it works with the pictures, titles and any qualifications.
Business.gov.au's AI guidance recommends human oversight and checking AI outputs before using or sharing them. For generated narration, make that practical: identify who listens to the actual final output, resolves unfamiliar names or terms and approves the version that enters the film.
Do not replace that responsibility with “the tool reads the script”. Equally, do not assume a recorded performance requires no checking. Define the relevant checks for either route and include their time and ownership in the comparison.
If a proposed tool would receive confidential script material, have the information owner assess that route before submitting it. Keep this separate from the creative decision about the voice. An attractive result does not settle the handling of the input.
Make revision promises specific
A proposal may present synthetic narration as easy to update. Ask what a real update would involve. Who changes the approved script, produces the revised audio, checks the whole affected passage and integrates it into the film? How are timing changes and related captions handled?
For recorded narration, ask about the corresponding recording and editing arrangement. Do not assume the narrator will remain available indefinitely or that an isolated new word will always fit seamlessly into an earlier performance.
For a synthetic route, ask what happens if the selected voice, service or applicable terms change. Receiving an approved audio file is different from being able to regenerate matching material later. Neither future availability nor unrestricted reuse should be inferred from the initial delivery.
Keep the approved track, script version and relevant use record identifiable. You are buying a usable production outcome, not merely access to the method that created it.
Compare the responsibilities side by side
| Buying question | Recorded performance | Synthetic narration |
|---|---|---|
| Authorised source | Who is engaged, for which recording and uses? | Which voice option and terms cover this use? |
| Direction | Who briefs, directs and selects the performance? | Who generates, adjusts and selects the output? |
| Verification | Who checks the delivered take in context? | Who checks the generated audio in context? |
| Revision | What recording and editing work would a change require? | What regeneration, checking and editing would it require? |
| Continuity | What can be confirmed about later recording arrangements? | What can be confirmed about later access to the same voice route? |
| Delivery | Which approved files and usage records are supplied? | Which approved files and usage records are supplied? |
Add the actual quoted scope beside the responsibilities. A low generation or recording charge alone does not describe the total work needed to release a finished film.
An illustrative proposal comparison
This example is fictional and does not describe a Live Images service or recommend a voice platform.
A business requests narration for a company explanation. One proposal includes a recorded performance, defined correction arrangements and delivery of the selected track. Another lists generated audio but leaves final checking and future revisions with unspecified people.
The buyer uses the comparison to make both proposals complete. It asks who would verify the generated output, what a revised passage would require and which voice-use terms apply. It also confirms the human narrator's actual usage and correction scope.
The team can then compare two workable routes. It does not conclude that either is cheaper or better from the headline line item, and it does not treat a convincing synthetic voice as evidence that a real customer or project participant made the statement.
Confirm the method before production relies on it
If the narration could be mistaken for a real person's account or endorsement, resolve that presentation before proceeding. Ask the relevant policy or legal reviewer about permissions and any disclosure needed for the actual use; no universal disclosure phrase is supplied here.
For a construction case study, discuss the intended explanation with Live Images and identify any narration method being proposed. Confirm suitability and scope explicitly. No synthetic voice, voice-cloning or standalone voiceover service is asserted as part of Live Images' current offering.
Prepared with AI assistance for Live Images. Planning guidance is general; suitability and production scope must be agreed for your project.