Choose the Right Voice Workflow for Your UGC Videos

Devon Ariza 17 August 2026

AI voiceover in a UGC video tool can mean several different things: choosing a preset synthetic voice, uploading a creator’s real recording, cloning an authorized speaker, translating an existing performance, or importing audio from a specialist voice generator. Those workflows solve different production problems, so there is no useful universal winner.

The better comparison is the complete path from script to publishable video. Assess voice and language suitability, revision effort, synchronization, commercial rights, plan restrictions, and cost per usable take—not just the most flattering demo or largest advertised voice library.

This guide is based on published product comparisons and vendor descriptions rather than independent hands-on testing. Capabilities, prices, quotas, and plan gates can change. Treat the product details below as a preliminary shortlist for verification, not a definitive current specification sheet.

The short answer: match the voice workflow to the UGC job

Start with the type of video you need to produce:

For multilingual avatar localization: Shortlist HeyGen. Published comparisons associate it with built-in voices, cloning, translation, dubbing, and avatar lip-sync. Reported language totals differ depending on whether the source is counting speech, translation, or dubbing, so verify the exact language and plan combination you need.
For a specialist voice library inside a talking-head ad workflow: Consider Arcads. One commercially interested comparison reports that it uses ElevenLabs and says cloning may require the customer’s own ElevenLabs account. That dependency could affect cost, permissions, and administration.
For product-page-to-ad production: Consider Creatify. It is positioned around turning product URLs into ads with scripts, presenters, and narration. Published comparisons place cloning on a higher plan rather than the entry tier.
For editing around a creator’s real recording: Consider Captions/Mirage as an audio-first candidate. The evidence conflicts: one comparison describes supplied audio and external cloning, while another reports cloning on a Business plan. Verify the current workflow before purchasing.
For business, training, or corporate-avatar localization: Consider Synthesia. Its broad multilingual positioning may suit structured explainers and training, but that does not establish a fit for deliberately casual, handheld creator-style UGC.
For maximum control over isolated narration: Test ElevenLabs alongside the video platforms. A specialist engine may offer more control over voice selection, pronunciation, cloning, and delivery, but importing and resynchronizing audio can offset that flexibility.

A competitor-published comparison associates Arcads with an ElevenLabs-dependent workflow, Creatify with URL-to-video ads, MakeUGC with multilingual text-to-speech, and Captions/Mirage with supplied audio. It does not provide a controlled listening benchmark across those products, so its specifications are useful for shortlisting rather than ranking quality. Review the reported UGC voiceover comparison.

Most importantly, the available evidence does not establish which UGC video tool sounds best. There is no controlled comparison covering the main integrated platforms with identical scripts, comparable settings, multiple languages, and the full revision process. Any definitive naturalness ranking would therefore be unsupported.

Five voiceover workflows that are easy to confuse

“AI voiceover” is an umbrella label. Before comparing products, identify which workflow a feature page actually describes.

1. Native text-to-speech

You enter a script, select a preset synthetic voice, and generate speech inside the video tool. The same environment may connect that speech to an avatar, captions, stock footage, or a social-video template.

This is often the fastest workflow for producing many product-ad variants. Its limitation is that preset voices may offer less control over unusual pronunciations, distinctive delivery, or a recognizable creator identity.

2. Uploaded human audio

A creator records a real performance and uploads the file as the spoken track. The platform may then add captions, edit the footage, clean the recording, synchronize visuals, or dub the content.

Uploaded audio is not synthetic voice generation. It can be preferable for testimonials, sensitive claims, emotionally specific scripts, recognizable creator delivery, or concepts where automation would weaken the intended authenticity.

Pictory is reported to support both AI voices and uploaded user audio, while InVideo AI generates narration within a prompt-to-video process. These are examples of integrated content-production workflows, not evidence that either has superior UGC voice quality. Compare their reported narration workflows.

3. Native voice cloning

The platform creates new speech intended to resemble an authorized speaker using samples supplied for that purpose. A clone can help a founder or creator maintain continuity across many scripts without manually recording every line.

“Native” matters. It means the video platform itself provides or directly manages the cloning workflow. It should not be confused with generating speech in another service and uploading the resulting file.

A clone is not automatically a complete substitute for the speaker. Cadence, emphasis, pronunciation, emotion, and consistency still need testing. Authorization, licensing, sample retention, and permitted uses require separate review.

4. External-engine narration

The spoken track is generated in a specialist service and then imported into—or connected with—the UGC video platform. Arcads paired with ElevenLabs is one reported example; manually generating an audio file in a specialist service and importing it into an editor is another.

This arrangement can provide deeper audio controls or broader voice selection. In return, the team may have to manage two subscriptions, transfer files, repair timing, rebuild captions, and repeat synchronization after script changes.

5. Dubbing and translation

Dubbing adapts an existing performance into another language. Depending on the product, the process may include translation, generated speech, timing adjustment, preservation or recreation of the original voice, and avatar or facial lip-sync.

These are separate capabilities:

Capability	What it does	What it does not prove
Speech generation	Converts written text into spoken audio	Translation accuracy or lip-sync
Translation	Converts meaning between languages	Natural pronunciation or convincing delivery
Dubbing	Replaces or adapts an existing spoken track	That the same languages support text-to-speech
Captions	Displays speech as text	That spoken localization is available
Translated captions	Localizes on-screen text	That the audio is translated
Avatar lip-sync	Aligns visible mouth movement with speech	That the underlying voice sounds natural

A claim of “100 languages” might refer to captions in one product, generated speech in another, and video translation in a third. Keep separate comparison columns for speech, captions, translation, dubbing, and lip-sync.

Capability matrix: what the leading options reportedly support

This matrix is a preliminary research checklist, not a verified current feature table. It summarizes what the supplied comparisons report and deliberately marks unresolved fields as unclear. “Not documented” means the cited evidence does not establish the capability; it does not prove that the product lacks it.

Tool	Primary workflow	Preset text-to-speech	Uploaded audio	Native cloning	External voice dependency	Generated-speech languages	Dubbing or translation	Lip-sync	Important verification note
HeyGen	Multilingual avatar speech and localization	Reported	Not documented	Reported	Not documented for the core workflow	Broad support reported; totals vary by feature definition	Reported	Reported	Verify whether the quoted total refers to speech, dubbing, or translation and which plan includes cloning.
Arcads	Talking-head UGC ads linked to a specialist voice library	Reported through the described workflow	Not documented	Not established as native	ElevenLabs reportedly used	Approximately 30–35 in one competitor comparison	Not documented	Reported	Cloning may require the customer’s own ElevenLabs arrangement; confirm inclusion and separate billing.
Creatify	Product URL to scripted video ad	Reported	Not documented	Reported on a higher tier	Not documented	At least 29 in one competitor comparison	Not documented	Not established by the supplied evidence	Confirm cloning tier, export allowance, and whether each target language supports the same features.
Captions/Mirage	Audio-first editing, talking-head video, and dubbing	Reported by one source	Reported	Conflicting reports	Possibly external for cloning	Not established	Multilingual dubbing reported	Reported for avatar workflows	Verify before purchase: sources disagree on whether cloning is external or available on a Business plan.
Synthesia	Business, training, and corporate-avatar video	Reported	Not documented	Custom voice options reported on higher plans	Not documented	Sources report broad coverage above 130 languages	Multilingual workflows reported	Reported	Do not assume a corporate presentation style will suit casual creator-led UGC.
MakeUGC	Multilingual UGC ad generation	Reported	Not documented	Reported on higher tiers	Not documented	More than 50 in one competitor comparison	Not documented	Not established by the supplied evidence	Confirm whether the language count refers specifically to generated speech and which tier enables cloning.
Pictory	Script or article repurposing with footage and narration	Reported	Reported	Not established	None required for the basic reported workflow	Not established	Not established	Not an avatar-first emphasis	Relevant to content repurposing, not proof of superior creator-style delivery.
InVideo AI	Prompt-to-video generation with narration	Reported	Not documented	Not established	None required for generated narration	Not established	Not established	Not the main reported capability	Test whether sentence-level audio changes preserve scene timing and captions.
Pippit	Social or commerce video creation	Reported	Not documented	Not established	Not documented	28 languages reported	Not documented	Not documented	A roundup reports more than 869 voices, but that quantity does not establish naturalness or market leadership.
Shhots AI	Product-ad generation with integrated audio	Reported	Not positioned as the core workflow	Not reported	Not reported	Its own site claims more than 12 languages	Not established	Integrated ad workflow reported	Its own site also claims five accent choices; verify availability in the intended plan.
ElevenLabs	Standalone speech, cloning, and dubbing	Reported	Not applicable as a UGC-video upload feature	Reported	It is the external engine	More than 70 reported	Reported	No native video generation in the cited comparison	Budget for importing, captioning, synchronization, and later script revisions.

The matrix relies heavily on vendor-authored comparisons, competitor comparisons, and third-party roundups. For example, Pippit’s reported 869-plus voices across 28 languages comes from a roundup focused more broadly on avatars, product presentation, and pricing—not a systematic listening test. The same source reports multilingual dubbing for Captions/Mirage. See the Pippit and Mirage feature report.

A vendor comparison reports Shhots AI’s own claims of more than 12 languages and five accent choices, while also positioning HeyGen for avatar localization, Synthesia for corporate video, ElevenLabs for standalone narration, Creatify for URL-led ads, and Captions for editor-based voiceover. These are workflow characterizations, not independently measured quality findings. Review the vendor-authored workflow comparison.

For Synthesia, supplied sources report different totals—above 130 in one roundup and above 140 in another. The defensible conclusion is that it is positioned as a broad multilingual business-avatar candidate. The exact language, voice, custom-voice, and plan combination still requires live confirmation.

Integrated UGC voiceover versus a standalone voice generator

An integrated UGC platform attempts to keep the production chain in one environment:

Enter or generate the script.
Choose a voice.
Select an avatar, template, stock scene, or uploaded footage.
Generate or align visible speech.
Produce captions.
Revise the script or voice.
Export a social-ready video.

The primary advantage is not necessarily better speech. It is fewer handoffs. When a team needs many hooks, aspect ratios, spokesperson variants, or localized versions, avoiding repeated audio import and synchronization can matter more than having the deepest voice controls.

A standalone workflow adds more steps:

Finalize the script and pronunciation notes.
Generate several narration takes.
Export the selected audio.
Import it into a video editor.
Synchronize scenes, cuts, and avatar movement.
Create or rebuild captions.
Correct timing problems.
Repeat the process after script changes.

That extra work may be worthwhile when delivery control is the bottleneck. A specialist engine can be preferable when the team needs detailed pronunciation handling, a reusable isolated audio file, an authorized clone used across several editors, API automation, or more choice over voice and performance.

ElevenLabs is the principal standalone contrast in the supplied comparisons. It is associated with broad voice selection, cloning, multilingual output, expressive controls, and developer access, but it does not provide the same end-to-end video synchronization described for integrated ad platforms. Those reports do not justify calling it objectively best-sounding. See the integrated-versus-standalone comparison.

Descript offers a middle path. Published comparisons describe it as combining audio and video editing, transcript-based corrections, and own-voice generation. That can be useful when the task is editing recorded media or replacing individual lines. It is not the same category as an avatar-first UGC generator that creates a synthetic spokesperson ad from a product page.

The best setup may therefore use more than one product. A brand might create avatars and social layouts in one tool, generate a difficult language or signature voice in another, and retain human recordings for sensitive testimonials. Procurement simplicity is useful, but it should not force every format and language through an engine that performs unevenly.

Choose integrated production when iteration speed is the bottleneck. Choose a specialist voice engine when delivery control or reusable audio is the bottleneck.

How to compare voice quality without trusting demo reels

Voice-library size, language count, avatar realism, and convincing lip movement do not establish narration quality. A voice can look synchronized while mispronouncing the brand name, flattening a disclaimer, or changing character between regenerated takes.

A useful testing framework is to run identical, difficult real-world copy through every finalist. Published voice-testing guidance recommends including names, numbers, acronyms, foreign words, warnings, and calls to action, then scoring both output quality and cost per usable take. Review the detailed testing framework.

Build a shared script containing:

A short, interruptive hook
Your product and brand names
A price
A date
An acronym
A foreign term or code-switched phrase
A factual disclaimer
A testimonial-style sentence
A direct call to action

For example:

Wait—before you spend $49 on the ACME Pro, here is what changed on 15 September. Our EU FAQ calls the feature mise à jour rapide, but results depend on your setup. I found the first step easier than expected, although your experience may differ. Read the full terms and choose your size today.

The copy is intentionally awkward. Easy demo sentences reveal little about how a system handles the friction found in real UGC advertising.

Run identical text through every finalist. Choose comparable voices, speeds, and styles; do not compare one platform’s carefully tuned premium voice with another platform’s arbitrary default. If a tool offers several plausible voices, test a small matched set rather than selecting the first option.

Score each output separately:

Criterion	What to examine
Pronunciation	Brand names, numbers, acronyms, foreign terms, and unusual product language
Pacing	Whether the hook moves quickly without rushing the disclaimer
Pauses	Natural phrase boundaries and intentional breaks
Emphasis	Whether benefits, qualifications, and calls to action receive appropriate weight
Emotional fit	Whether the delivery suits a testimonial, product demo, founder message, or warning
Accent fidelity	Whether the claimed accent sounds credible to native listeners
Consistency	Whether regenerated lines retain the same voice character and energy
Synchronization	Whether speech, mouth movement, cuts, and captions remain aligned
Edit effort	How much human intervention is needed before publication

Native-speaker review is essential for every target market. A supported-language total cannot show whether a regional accent is credible, a translation is idiomatic, or a product name sounds unintentionally comic.

Track production friction as data. Count:

Regenerations
Manual pronunciation changes
Translation corrections
Timing adjustments
Failed renders
Caption repairs
Full-scene or full-video rebuilds

Also test a sentence-level correction. Change one price, product feature, or disclaimer and see whether the platform can rebuild only that passage. If a two-second revision forces the team to regenerate the narration, captions, avatar synchronization, and final render, the apparent speed advantage may disappear.

One editorial comparison used the same 25-second excerpt across several standalone generators. That is a useful methodological example because it controls the copy, but its selected voices and short sample cannot establish performance across UGC platforms, languages, or longer campaigns. Review the limited shared-script test.

Where practical, conduct blind listening. Remove tool names, normalize playback volume, and ask reviewers to score files before seeing which platform generated them. Retain every output, settings note, and timing log so decision-makers compare actual production evidence rather than marketing terms such as “studio-quality” or “ultra-realistic.”

The final measure should include cost per usable take. A voice that wins the first listen but requires five repairs whenever a price or product name changes may be the weaker production choice.

Calculate the real cost per publishable UGC video

The headline monthly price is rarely a complete comparison. Plans may meter characters, credits, generated seconds, finished minutes, projects, downloads, or completed videos. Two subscriptions at the same advertised price can therefore produce very different amounts of commercially usable work.

Use this worksheet:

Effective monthly workflow cost =
monthly subscription fee OR amortized annual subscription fee
+ cloning or custom-voice upgrade
+ external voice-engine subscription
+ usage overages
+ paid export or storage costs
+ production labor
+ expected cost of discarded generations

Cost per publishable video =
effective monthly workflow cost
÷ number of videos that pass review and can legally be published

Do not add the monthly fee and the amortized annual fee for the same subscription. Use whichever reflects the billing arrangement you will actually purchase, then add only incremental costs.

If audio is the reusable deliverable, also calculate:

Cost per usable finished minute =
total voice and production cost
÷ approved, downloadable, commercially usable minutes

Do not put every generated second in the denominator. Exclude failed renders, rejected takes, duplicate tests, unusable translations, and output that cannot be downloaded or used for the planned commercial purpose.

Include revision waste

A script might require three attempts because the first mispronounces the product, the second rushes the disclaimer, and the third loses synchronization after a caption change. If the plan deducts credits for every generation, all three attempts are production cost.

Assign a labor value to:

Creating pronunciation spellings
Importing standalone audio
Trimming silence and correcting timing
Rebuilding captions
Rechecking translated copy
Synchronizing lip movement
Moving files between tools
Managing separate subscriptions and access permissions

A cheaper standalone voice subscription can become the more expensive workflow if an editor spends substantial time repairing every variation.

Separate allowance from usable output

“Ten minutes included” can mean ten minutes of generation, ten minutes of downloaded audio, or ten minutes before retakes. It may or may not include commercial rights. A free tool may permit voice previews while restricting downloads or commercial use.

Ask four distinct questions:

How much can we generate?
How much can we download or export?
How much survives revisions as usable output?
Which output is licensed for the planned publication and advertising use?

Reported by source—verify live

A competitor comparison published in July 2026 reported entry prices of about $29 per month for HeyGen, $39 for Creatify, $59 for MakeUGC, and approximately $110 for Arcads. It also reported higher-tier cloning for Creatify and MakeUGC and a possible separate ElevenLabs dependency for Arcads cloning. These are dated examples of plan structure, not a current pricing table. See the dated UGC-platform comparison.

Standalone plans create similar traps. A 2026 review reported that free output from several generators lacked commercial rights and that paid products differed in whether they metered credits, generated time, or downloaded minutes. It also reported that discarded ElevenLabs takes consumed credits under the plan reviewed. See the dated plan and licensing comparison.

A low entry price is not a bargain if it excludes the clone you need, sufficient revision capacity, downloadable files, commercial use, or paid-ad rights. Compare the configured workflow you will actually operate—not the cheapest logo on a pricing page.

Commercial rights, cloning consent, privacy, and disclosure

Before publishing sponsored UGC, paid ads, or client work, complete a rights and risk review. This is practical screening, not legal advice. Licensing terms and platform policies should be checked for the intended territory and use.

Pre-publication rights checklist

Confirm in current documentation or a written agreement:

Does the plan permit commercial use?
Are paid advertisements covered?
Are sponsored posts and monetized channels covered?
Can the output be delivered to a client?
Can isolated audio be reused outside the generated video?
Do rights differ between preset voices, cloned voices, and uploaded recordings?
Are there territory or duration limits?
Do free and paid plans grant different rights?
Does an upgrade apply to earlier files or only future output?
Can the same voice be used across several brands or accounts?
Are there restrictions on political, medical, financial, adult, deceptive, or other sensitive content?

Free generation or download does not by itself establish commercial-use rights. Published plan reviews report examples in which free output could be generated but was non-commercial, or trial audio could not be exported. Check the license that applied when the file was created rather than assuming a later upgrade changes earlier rights.

Cloning authorization

Obtain documented authorization before cloning a creator, founder, employee, customer, voice actor, or other person. The record should identify the speaker, intended uses, clients or brands, channels, territory, duration, and whether the voice can be reused after the relationship ends.

Consent is necessary risk control, but informal permission does not settle every contractual or legal issue in every jurisdiction. Review applicable agreements and seek professional advice when the stakes justify it.

For stock voices, ask how recordings were sourced, what performers authorized, how they are compensated, and whether the vendor’s license covers your planned commercial use. For uploaded samples, ask whether recordings can be retained, reused, shared with subprocessors, or used for model improvement.

Privacy and security

Teams entering unreleased product claims, customer information, financial details, or internal scripts should ask:

How long are scripts, recordings, and voice samples retained?
Can customers delete them?
Where are they processed and stored?
Are they used for model training?
Are data encrypted in transit and at rest?
Which employees and subprocessors can access them?
Are role-based permissions, single sign-on, or audit logs available?
Can administrators trace who created or exported a clone?

Security certifications can indicate that specified controls have been audited. They do not make a provider universally compliant, suitable for every dataset, or risk-free. A vendor-authored risk guide recommends examining voice sourcing, licenses, retention, processing locations, encryption, permissions, and auditability; its claims about the vendor’s own service remain self-reported. Review the AI voice licensing and security checklist.

YouTube disclosure

YouTube’s first-party guidance lists cloning your own voice for voiceovers or dubs as an example that generally does not require disclosure. By contrast, realistic synthetic audio that makes another person appear to say something they did not say requires disclosure. Creators can use the “AI use” setting in YouTube Studio, and disclosed content may receive an altered or generated label. Read YouTube’s GenAI disclosure guidance.

Those examples are non-exhaustive. The decision depends on realism, the significance of the alteration, and whether viewers could be misled. The own-voice example does not authorize cloning another person, and the supplied evidence does not establish equivalent rules for TikTok, Instagram, or other platforms. Check each destination’s current policy separately.

Decision guide: preset voice, own-voice clone, human recording, or specialist engine

Choose the voice source only after defining what the audience needs to believe, understand, and feel.

Use preset text-to-speech for rapid variants

Preset voices are a sensible starting point when:

The spokesperson’s identity is not central
The campaign needs many hooks or product variations
Scripts change frequently
Speed matters more than distinctive delivery
The selected voice passes the shared-script test

This approach can suit straightforward product features, offer variations, and temporary creative tests. It is less suitable when the concept depends on a known creator’s personal recommendation.

Use an authorized own-voice clone for continuity

An own-voice clone can fit founder-led or creator-led campaigns that need many scripts, revisions, or languages while preserving a recognizable identity.

Proceed only after verifying authorization, commercial rights, retention, model-training terms, language performance, and plan requirements. Test whether the clone remains consistent across emotional reads and regenerated lines; resemblance alone is not enough.

Use a human recording when performance is the asset

Record the person when nuance, credibility, emotional sensitivity, or authentic creator delivery matters more than automation. Human audio may be the strongest choice for personal testimony, apologies, complex claims, vulnerable topics, humor, or distinctive founder storytelling.

AI can still assist with cleanup, transcription, captions, translations, and editing around that recording. Choosing human delivery does not require rejecting automation elsewhere in the workflow.

Use a specialist engine when audio control is the bottleneck

A standalone generator is appropriate when the team needs:

Detailed delivery or pronunciation control
Reusable isolated audio
The same narrator across several editors
API-based production
A specialist cloning workflow
Voice output not tied to one avatar platform

Budget for synchronization and revision labor. If every script change requires a new export, import, caption pass, and visual retime, document that cost before scaling.

Use multilingual avatar tools only after local review

Do not approve localization from a language-count claim. Ask native speakers to review translation, pronunciation, accent, tone, timing, cultural fit, captions, and lip-sync for every target market.

One engine may perform well in one language and poorly in another. Using more than one provider by language or format can be more defensible than accepting uneven output merely to simplify procurement.

Build the shortlist in five steps

Define the job. Decide whether you need fast preset narration, an authorized creator clone, a real performance, localized avatar speech, or reusable specialist audio.
Verify capabilities and plan gates. Separate speech, captions, translation, dubbing, cloning, uploads, exports, and lip-sync.
Run identical scripts. Use difficult real copy and comparable voice settings.
Calculate cost per usable take. Include failed generations, retakes, synchronization, labor, upgrades, and external subscriptions.
Review rights before publishing. Confirm commercial use, paid advertising, client delivery, cloning authorization, retention, and disclosure requirements.

Do not choose a UGC voiceover tool from a demo reel, headline price, or language count. The most defensible choice is the workflow that produces publishable, properly licensed output for your team’s actual scripts—not the platform with the broadest marketing claim.

Frequently asked questions

Which UGC video tool has the most natural AI voice?

The available evidence does not establish one. Existing comparisons use different scripts, voices, settings, languages, and product categories; many do not conduct controlled listening tests.

Run blind, shared-script tests with comparable voices and score pronunciation, pacing, emotional fit, consistency, synchronization, and revision effort. The winner may vary by language, script type, and selected voice.

Which UGC tools let me upload a real voice recording instead of using text-to-speech?

Pictory is reported to support uploaded user audio. Captions/Mirage is also described as suitable for supplied voiceover in an audio-first workflow, although its current cloning arrangement is disputed by the supplied comparisons.

Other editors may support audio import even when it is not their headline feature. Confirm file formats, synchronization behavior, export rights, and whether replacing one line forces a full rebuild.

Is voice cloning included in entry-level plans?

Do not assume so. Published comparisons place cloning on higher tiers for Creatify and MakeUGC, while custom voice options may require higher Synthesia plans. Arcads cloning may depend on a separate ElevenLabs arrangement. Reports conflict over whether Captions/Mirage cloning is external or available on a Business plan.

Verify the current plan, required sample process, commercial rights, output allowance, and every external dependency before purchasing.

Can I use free AI voiceover output in paid ads or sponsored UGC?

Not automatically. Free access may permit generation or previewing while withholding commercial rights, downloads, or paid-ad usage. Do not assume that upgrading later retroactively licenses files created under earlier terms.

Confirm paid advertising, sponsored content, client work, monetization, territory, duration, and isolated-audio reuse in the terms that apply to the specific output.

Does YouTube require disclosure when I use an AI or cloned voice?

It depends on the use. YouTube lists cloning your own voice for voiceovers or dubs as an example that generally does not require disclosure. It requires disclosure for realistic synthetic content that makes another person appear to say something they did not say.

The examples are not exhaustive. Consider the realism and significance of the alteration and whether viewers could be misled, then use YouTube Studio’s disclosure setting when required.