Voice and publish

Creators use this path when the paid bottleneck is the voice. The export is a script a person can stand behind, an ElevenLabs narration, dub, or clone of their own voice, an optional picture in CapCut, and a disclosure check before the file goes to YouTube, TikTok, or YouTube Shorts.

Platforms: YouTube · TikTok · Shorts

This is the voice track and a light publish finish. The full generate stack, with a shot plan in the monthly sum, stays on the faceless channel workflow. A face stays on AI avatar. ElevenLabs and CapCut stickers below are the ones on those tool pages, verified 2026-10-16. HeyGen dollars, where they appear, are the reading already on AI avatar. This page does not estimate views or revenue. Search Console volume was not observed. A Google Trends index was not exported, so relative interest was not observed. TikTok Creative Center was not opened while signed in, so a topic score, an ad-library count, and any voiceover template ID were not observed. Noted queries in that pass were elevenlabs youtube, ai voiceover shorts, text to speech tiktok, and voice clone commercial use. Competitor lists in the same pass were vendor pages and "best AI voice" roundups. Murf is listed, verified 2026-10-17, and its sticker stays on that page, out of this sum. PlayHT and Speechify are still not listed. play.ht closed the connection and speechify.com returned a bot challenge on 2026-10-17, so those prices are not observed.

Who it is for

You already have a narration job, or you are about to have one, and the line you expect to pay is the voice. A faceless narration, a translation dub of a source you can distribute, or a clone of your own voice all land here. You do not need a default face. A seconds-per-word rate was not observed. This site lists no script product. Write the words in any editor, then start paying at the voice. Someone else's upload is not a dubbing source.

Recommended path

Four steps. The picture stays optional. The plan ladder lives in the cost section. Each step names the tool, the job, and why it is in the path.

1. Script

Write sentences you can stand behind. ElevenLabs reads them. It does not decide them. A draft from a chat product is still your edit before it becomes audio. Regenerating speech spends credits. The FAQ on the ElevenLabs pricing page, as the tool page records it, puts V2 multilingual text to speech at about 1 credit per character, so a rewrite is a new character count, not a free pass. Flash and Turbo on the API can cost 0.5 to 1 credit per character. What one website button spends above that rate was not observed. A full credit table for Eleven v3 and Eleven v4 was not observed. The pricing page names Eleven v4 as a current emotive model and lists v4 and v3 on the compare table. Treat 1 credit per character as the V2 multilingual figure, not as every model.

For a Short or a TikTok, write the length you will actually publish. The ElevenLabs tool page uses a 30-second Shorts narration as a small slice of the Creator grant. That sentence is an illustration of scale. It is not a quota, and it is not a words-per-minute rate. A dub starts from a source you can distribute, in the language you are leaving, not from a transcript of a video you do not hold.

2. Voice · ElevenLabs

ElevenLabs is this step because the job is the track: text to speech, an instant or professional clone, or a dub. It does not generate the picture this directory publishes. CapCut premium text to speech is a separate credit on CapCut's help pages, and it does not replace a cloned ElevenLabs voice. HeyGen includes voices because its product is a person on screen. Use it when you need a mouth. Which HeyGen plans accept an external audio file was not observed on the AI avatar page. If you do not need a mouth, skip HeyGen and keep the ElevenLabs track.

The tool page's source is elevenlabs.io/pricing, monthly view, checked 2026-10-16. A separate API price page at elevenlabs.io/pricing/api was not observed. Do not mix an older characters-per-dollar screenshot into these stickers.

Free is the draft card. It has no commercial-license line and no rollover, and it lists 3 studio projects. Starter is the first card with that line, plus instant voice cloning, music commercial use, dubbing studio, image and video, and 20 studio projects. The image-and-video row is not the picture on this path. The tool page says ElevenLabs does not generate the frame this directory publishes. Budget the renewal in the cost table, not the first-month price. The license agreement was not opened, so the exact commercial grant is not observed. The plan-card label is the line. It is not a paraphrase of the contract. Creator is the card that adds professional voice cloning, and it replaces Starter. It is not a second subscription on top of the floor.

Credits are one pool, so a music bed competes with the read. Dubbing rates, rollover, and the plans above Creator are in the cost section. A single-request credit cap was not observed. It depends on free versus paid.

3. Picture (optional) · CapCut

Skip this step when the export is podcast-style audio and you are not burning captions or assembling a frame. There is no CapCut line on that export.

Open CapCut when the voice still has to become a YouTube video, a TikTok, or a Short: trim, reframe to 9:16, burn captions, export. Keep 16:9 when the YouTube video is long. ElevenLabs stops at the track. The timeline is why CapCut is the optional picture. A measured pixel inset was not observed. Type, a safe zone, or a cover that still fails continues on captions and covers. The pricing page did not print a cover spec.

The CapCut tool page, checked 2026-10-16, does not print a free-tier limit table. Watermark, resolution, and caption quotas on a free export were not observed there. Captions and covers records a help note dated Apr 3, 2026: auto captions can include Pro-only functionality, a Standard user may preview, and an export that contains Pro elements requires Pro. Standard is not a paid tier on the tool page. This page did not re-open that help article. Credits pay for advanced AI use, including premium text to speech, and they cannot buy the plan. A credit integer for that CapCut speech was not observed.

Stickers, grants, and Seedance ceilings are in the cost section. Pay the currency on screen. Do not add a CLP or yen card to the voice floor. Seedance is the model family on the paid card. A shot under every sentence belongs on the faceless channel workflow.

4. Publish and disclose

Upload after the label matches the file. This step has no subscription. This is not legal advice. The labels below are the ones already recorded on the faceless channel workflow, AI avatar, and UGC ads. This page did not re-fetch the HTML. They do not clear an upload.

The YouTube help page those workflows cite, https://support.google.com/youtube/answer/14328491?hl=en , is the page whose examples include cloning your own voice for narration among the items that do not require disclosure. The same reading lists non-realistic scenes, script or title help, and captions on that side of the line. The control they name is YouTube Studio, then Attributes, then AI use, then Yes or No. A realistic AI face is not the narration example. A photoreal avatar, including a consented twin, still uses the AI-use control, and that file belongs on AI avatar. Photoreal b-roll of a fictional event still uses the toggle even when the narrator is your own clone. A mixed file still discloses the realistic part. On that help page, as those workflows record it, disclosure itself does not limit audience or monetization eligibility.

On the faceless workflow's reading, TikTok requires a label for realistic AI images, audio, and video, and it encourages one for fully generated or heavily edited AI. That heavily edited case is what this path means by a significant AI change. AI avatar cites the 19 September 2023 newsroom post for the realistic-media label and does not treat an Ads Manager checkbox as observed. This page does not add a template ID. A signed-in Creative Center topic score was not observed.

A voice that is not yours is not the YouTube narration example. Another person's face or voice, without agreement, is the refusal already stated on AI avatar. A label does not replace that agreement. Instant cloning is the Starter card. Professional cloning is the Creator card. What those two labels require in sample length was not observed beyond the card text.

Monthly starter cost

Checked against the tool pages dated 2026-10-16. Tax sits outside the quote. Assumption: one English narration, or one dub of a source you can distribute, voice from ElevenLabs. If there is a picture, Shorts and TikTok use 9:16 and a long YouTube video can stay 16:9. Compare-table minutes wear a tilde. No post count, because a character-to-video conversion depends on script length and on which model you run, and a single-request credit cap was not observed. No view or revenue figure. HeyGen, Ultra, Teams, ElevenLabs Pro, Scale, and Business are not in the four bands.

Band	What you pay for	Sticker	What it is not
Draft	ElevenLabs Free. CapCut only if you choose a picture, and then without Pro	$0. 10,000 credits, about 10 minutes. Free watermark, resolution, and caption quota not observed	A commercial-license line, or rollover
Commercial floor	ElevenLabs Starter, monthly renewal	$6 (30,000 credits, about 30 minutes). First month $1, labeled until 18 October 2026	The $1 as the plan price. The contract text was not observed
Floor + USD CapCut Pro	Starter renewal plus the USD Pro card	$6 + $19.99 = $25.99 before tax	CLP 29990, or JPY 2,700, added to $6
Professional clone	ElevenLabs Creator, monthly renewal	$22 (121,000 credits, about 121 minutes). First month $11	$6 plus $22. Creator replaces Starter

The default branded month is $6. It is $25.99 before tax when you also need a picture and CapCut shows the USD Pro card. In the Starter promo month the voice side is $1, so the same USD card is $1 + $19.99 = $20.99 before tax for that month only, then $25.99. CapCut's $19.99 is not a first-month subscription discount.

Creator replaces the $6 floor when the professional clone, or the compare-table minutes, are the limit. Creator plus the same USD card is $22 + $19.99 = $41.99 before tax, or $11 + $19.99 = $30.99 in Creator's first month. Both figures are arithmetic, not a fifth band, and not a bundle. Do not add them to $25.99. That would count Starter and Creator together.

The tool page's other rendering is CLP Pro 29990, Ultra 71990, and Teams 14995 per seat, with the same grants as USD: 1,200 on Pro and Teams, 8,645 on Ultra. Pay CLP beside the ElevenLabs dollar. Do not fold it into $25.99. The tool page prints no yen. The faceless channel workflow records a same-day yen open: Pro JPY 2,700 with 800 credits, Ultra JPY 12,600 with 13,821, and the 2-seat Teams card JPY 1,295 per seat with 800. Long-to-short, UGC ads, and AI avatar print that same yen Pro sticker. Leave every yen sticker out of the dollar sum. Do not read 8,645 as the yen Ultra grant. USD Ultra at $79 replaces Pro when the second ceiling is the limit. USD Teams at $12.49 per seat is the shared editor. Neither is the voice floor.

Pro's Seedance 2.0 Mini ceiling is up to 240 seconds at 720p or 600 seconds at 480p. Ultra's is up to 1,700 seconds at 720p or 4,300 at 480p. Those ceilings are tied to the grant. They are not a narration, and they are not a promise that every prompt returns at that length. First-month lines on the Pro card, and on the Teams card per seat, start at Seedance 2.5 from $0.12 a second and Seedance 2.0 Mini at 480p from $0.007 a second. Ultra's first-month lines are Seedance 2.5 from $0.18 a second and Seedance 2.0 Mini at 480p from $0.009 a second. The rate after month one was not observed. Storage is stated as 1 TB on the tool page for Pro, Ultra, and Teams. A 1 TB line on the yen card was not observed. A public CapCut API and an official MCP endpoint were not observed. The Teams card said a signed-in visitor could save up to 80 percent. That discount was not itemized, so the saved amount is not observed.

Dubbing does not inherit the compare-table minutes. Those minutes match the text-to-speech estimate. A music bed is about 900 credits a minute on the pricing FAQ, so it spends the same pool as the read. Watermarked automatic dubbing is 2,000 credits a minute. Clean automatic dubbing is 3,000. The tool page's worked example is about 40 minutes of that clean mode on the Creator grant. The same rate against the 30,000-credit Starter grant is 10 minutes. That 10 is arithmetic from the FAQ rate and the grant, not a separate vendor sentence. Dubbing Studio is 5,000 with a watermark and 10,000 without. Extra minutes on the compare table were about $0.20 on Starter and about $0.18 on Creator. Across the whole table the extra-minute figures were about $0.36, $0.20, $0.18, $0.17, $0.17, and $0.17. A top-up pack price was not observed. Top-up credits are a separate balance and are not under the rollover cap.

Paid rollover lasts up to two months, up to 2 times the monthly quota, so the balance can reach at most 3 times the monthly quota while you stay on that paid plan and do not downgrade or cancel. Downgrading or canceling forfeits unused credits at the end of the cycle. Free does not roll.

Above Creator, Pro is $99 for 600,000 credits. The card lists 44.1 kHz PCM via the API. 192 kbps is on that card and, on the compare table, from Creator upward. The 192 kbps row is not the definition of a professional clone. The compare table's minute estimates continue at about 600, 1,800, and 6,000 on Pro, Scale, and Business. Scale is $299 for 1,800,000 credits, with 3 seats and 3 professional voice clones. Business is $990 for 6,000,000 credits, with 10 seats, 10 professional voice clones, and low-latency text to speech as low as 5 cents a minute. Enterprise is custom. A public Enterprise dollar price was not observed. The API dollar table was not observed, so no per-character API dollar is in this sum. Pro replaces a cheaper voice card. It is not added to $6. An official hosted MCP server at https://api.elevenlabs.io/v1/mcp can return speech audio as a short-lived download link. Dubbing, music, and speech to text were not in the MCP tool list that was read.

Annual billing, from the FAQ, is 10 months' price. The equivalent monthly figures are $5 for Starter, $18.33 for Creator, $82.50 for Pro, $249.17 for Scale, and $825 for Business. Those are FAQ figures, not a checkout this directory completed. CapCut's annual sticker was not observed, so those equivalents stay off the $25.99 line. The compare table, not the FAQ, lists 74 languages on each column, custom voice slots of 3, 10, 30, 160, 660, and 2,200, and audio quality of 128 kbps at 44.1 kHz on Free and Starter, with 128 and 192 kbps at 44.1 kHz from Creator upward. Sample length for either clone label was not observed.

Staged alternatives

Move when the limit named here is the one you hit. Dollar stickers stay in the cost section. Each stage has three alternatives.

Voice

Leave Free before a branded publish. The commercial-license line starts on Starter, and the contract behind it was not observed.

• ElevenLabs Free. Drafts. No rollover and no commercial-license line.
• ElevenLabs Starter. The commercial floor. Instant voice cloning and dubbing studio start here. A clean automatic dub is a much shorter slice than the compare-table text-to-speech minutes.
• ElevenLabs Creator. Replaces Starter for a professional clone, or when the Starter minutes run out. It is not Starter plus Creator. Sample-length rules for the clone were not observed.

Pro is the PCM and larger-grant card. Scale and Business are the seat cards. Enterprise has no public dollar on the extract. None of those four is the floor.

Optional picture

• Skip CapCut. Podcast-style audio, no captions, no frame. The month stays on the voice card.
• CapCut without Pro. Cuts and captions when a frame is required. Watermark, resolution, and a caption quota were not observed. An export that contains Pro elements is the captions-and-covers caveat in the picture step.
• CapCut Pro. The USD picture line in the $25.99 band. CLP and yen Pro are the same job in other currencies. Do not add the cards.

Ultra replaces Pro when the Seedance second ceiling is the limit. It stays out of $25.99. Teams is a shared draft, not a second narrator. No second caption app sits on this path. CapCut's own premium text to speech does not replace the ElevenLabs clone.

Publish

• YouTube, own-voice narration. The help-page example that need not use the AI-use toggle. Script help and captions sit on that same example list. A realistic face, or photoreal footage of a scene that was not filmed, still uses the control.
• TikTok. On the faceless reading, realistic AI images, audio, and video require a label, and fully generated or heavily edited AI is encouraged. No template ID was observed. Ads Manager clicks were not observed.
• Captions and covers. When the track is fine and the type, the safe zone, or the cover is not. That handoff adds no voice charge.

Where it fits

• The paid line is the voice: text to speech, a dub, or a clone of your own voice, from a script you can stand behind.
• You will publish under a brand only after you have read the commercial-license line on Starter, and you will keep Free for drafts. The contract text was not observed, so the card label is as far as this page goes.
• The picture is optional. A podcast-style audio export skips CapCut. A Short, a TikTok, or a narrated YouTube video opens CapCut for the frame and the captions.
• You will price a dub at the per-minute dubbing rates in the cost section, not at the compare-table text-to-speech minutes.
• Delivery is YouTube, TikTok, or Shorts, with the disclosure split in the publish step: own-voice narration on one side, a realistic face on the other.

Where it does not fit

• A talking face, a photo avatar, or a Digital Twin. That is AI avatar and HeyGen. Audio import into HeyGen was not observed by plan.
• A product URL in and a vertical actor ad out. That is UGC ads.
• Shots under every sentence, plus the voice, as one monthly stack. That is the faceless channel workflow. This page does not add a generator sticker.
• Only captions remain, on a file that already has the right voice. That is captions and covers.
• The job is to find highlights in a long file you already have the right to cut, and to keep that recording's speaker. That is long-to-short.
• The voice or the face is someone else's, without agreement. The YouTube narration example is your own voice. A label does not replace agreement.
• You want a view estimate, a revenue estimate, a Trends index, or a TikTok template ID. Those were not observed.
• You want a Murf, PlayHT, or Speechify dollar. Murf has that sticker, verified 2026-10-17. PlayHT and Speechify are still not listed. play.ht closed the connection and speechify.com returned a bot challenge on 2026-10-17, so those two prices are not observed.

Pitfalls

The $1 treated as the plan. Starter's first month is a promo in front of the renewal. Creator's first month is the same kind of teaser. Annual FAQ equivalents are not the monthly band. CapCut's USD Pro sticker is the subscription in the region that showed it, not a $1 add-on. The numbers are in the cost table.

A Free voice on a branded publish. The commercial-license line starts on Starter. The contract was not opened. Keep Free for drafts. Do not paraphrase a license the card does not print.

CLP or yen folded into $25.99. The USD picture month is Starter's renewal plus the USD Pro card. CLP keeps the 1,200-credit grant. Yen Pro, from the faceless page, changes the grant to 800. Ultra's USD card replaces Pro. It does not sit on top of the floor as a second editor.

Dubbing priced as narration. Compare-table minutes are text-to-speech estimates. Clean automatic dubbing spends the shared pool much faster, and a music bed spends the same pool. Rejected takes spend it again. The per-minute rows are in the cost section.

The narration example applied to a face. YouTube's skip-disclosure example is your own voice for narration. A realistic AI face is a different case. Photoreal b-roll of a scene that was not filmed still uses the AI-use control. On the faceless reading, TikTok requires a label on realistic AI media.

HeyGen, or CapCut speech, added quietly. On the AI avatar page, HeyGen Creator is $29 a month, or $24 per month if paid annually, with 600 credits that are not a minute count. Free is up to 3 videos a month at 1 minute. Which plans accept external audio was not observed, so those dollars stay out of $6 and out of $25.99. CapCut premium text to speech does not replace the clone you are paying ElevenLabs for.

Next workflow

Phone-readable captions, safe zones, and a cover that still fail go to captions and covers. A shot under the narration, as part of the monthly stack, goes to the faceless channel workflow. A consented face goes to AI avatar and HeyGen. Confirm audio input on that plan before you assume the ElevenLabs file drops in. A product-URL actor ad goes to UGC ads. A long file you will only cut, keeping its speaker, goes to long-to-short.

FAQ

How much is the starter month, in one number?

$6 for ElevenLabs Starter at renewal, or $25.99 before tax when the USD CapCut Pro card is on the same month. Both were on the tool pages dated 2026-10-16. Creator replaces Starter. It is not added to $6. The first-month promo and the rest of the ladder are in the cost table.

Is the $1 price the plan?

No. The card showed $1 for the first month and $6 as the price, labeled until 18 October 2026. Budget the renewal. A later open may not show the $1. CapCut has no matching first-month subscription price on the tool page. Its Seedance per-second lines are first-month rates, and the renewal rate was not observed. Annual equivalents are FAQ figures in the cost section, not this band.

Does my own voice clone skip YouTube's AI disclosure?

For narration, yes. Cloning your own voice is on the help page's list of examples that need not be disclosed. The page is https://support.google.com/youtube/answer/14328491?hl=en , as already used on the sibling workflows. This page did not re-fetch it. The control is in the publish step. A realistic AI face is not that example. On the faceless reading, TikTok still requires a label on realistic AI images, audio, and video.

Can the free plan publish under a brand?

Use Free to hear a voice. It has no commercial-license line. A branded publish on this path starts at Starter, where the line is printed. The contract text was not observed. Paid credits can roll for up to two months, inside the cap in the cost section, and downgrading or canceling forfeits what is left.

How much dubbing is in the Creator grant?

About 40 minutes of clean automatic dubbing, which is the tool page's worked example for the Creator grant. Compare-table minutes are the text-to-speech estimate, not that mode. The other per-minute rows, and the 74-language count, are in the cost section. That language count is not a promise that every dubbing mode matches the narration rate.

Do I add the CLP or yen CapCut sticker to $25.99?

No. $25.99 is the USD pair only. CLP Pro keeps the 1,200-credit grant. Yen Pro on the faceless page is 800 credits, not that grant. Pay each checkout in its own currency. The stickers, including Ultra, are in the cost section. Ultra is not a line on the voice floor.

Can I drop this track into HeyGen?

Not observed. The AI avatar page says HeyGen's comparison had an audio-input row, and which plans include it was not a clear yes or no. HeyGen is the mouth, not the default. Its Free and Creator stickers stay off $6 and off $25.99. If the avatar plan will not take the file, publish HeyGen's voice on AI avatar, or publish the ElevenLabs track with no face.

Does the hosted MCP replace a dub?

No. The official server at https://api.elevenlabs.io/v1/mcp can generate speech audio as a short-lived download link. Dubbing, music, and speech to text were not in the tool list that was read. Dubbing stays on the subscription, at the per-minute rates in the cost section. The API dollar table was not observed. PCM at 44.1 kHz via the API is a Pro card feature, and that plan replaces a cheaper voice card.