Skip to content
Dubsheet
Sign inSubmit a tool

Buyer's guide Text to speech and AI voice-over

Choosing an AI voice generator: meters, rights, cloning

Pick an AI voice by the unit its price counts, what the free plan leaves on your audio, commercial rights and whether cloning needs consent.

Dubsheet ResearchPublished 18 min readEditorial policy
See the comparison tableMarkdown
On this page
  1. The short answer
  2. The 7 tools compared
  3. What will an hour of finished speech cost me each month?
  4. What does each plan really count: characters, minutes or credits?
  5. What does the free plan forbid or leave on my audio?
  6. Can I publish the audio commercially, and who owns it?
  7. What happens when I run out of characters or minutes?
  8. Can I clone my own voice, and will it train their models?
  9. The tools that stand out
  10. Prices compared
  11. Which one to pick
  12. How we compared
  13. Questions buyers ask
  14. Sources

The short answer

Voicemaker at $5 a month and Typecast at $5 a month both grant commercial rights on that entry plan. Speechma and TTSMaker go further and allow commercial publishing from their free plans, with no credit required on the audio.

Fish Audio fits two hours of speech a month into its $15 a month Plus plan, using 120 of its 200 minutes, and Murf AI adds dubbing of your audio or video into other languages.

On 27 October 2026 we listed 8 tools: 7 of them under Text to speech and AI voice-over and 7 with a free plan.

  • Narakeet sells minutes in packs, paid once.
  • Typecast starts at $5 a month, the same entry price as Voicemaker at $5 a month.
  • Free plans are the norm, with 6 of the 7 offering one.
  • Buying more usage mid-cycle works on Narakeet, TTSMaker and Voicemaker, the 3 of the 7 that sell extra past the allowance.
Logos of Fish Audio, Murf AI, Narakeet, Speechma, TTSMaker, Typecast, Voicemaker. 7 tools compared, from $5 a month.

The 7 tools compared

Table 1 Price, free plan and the questions that separate them, checked 26 October 2026
ProductFromFree planCaptions and subtitlesPromotion and regular pricePast your allowance
Fish Audio
Free, then $15 a month
Yes
Partly
Yes
–
Murf AI
$29 a month
–
Yes
Yes
Partly
Narakeet
Free, then $6 per 30 minutes
Yes
Partly
No
Yes
Speechma
Free, then $9 a month
Yes
Yes
No
No
TTSMaker
Free, then $13.99 a month
Yes
–
Yes
Yes
Typecast
Free, then $5 a month
Yes
Partly
–
Partly
Voicemaker
Free, then $5 a month
Yes
Yes
No
Yes

What will an hour of finished speech cost me each month?

Fish Audio fits two hours of speech a month into its $15 a month Plus plan, using 120 of its 200 minutes, and Plus is its cheapest tier that allows commercial use. Typecast and Voicemaker open at $5 a month, Speechma at $9 a month and Fish Audio at $15 a month. Narakeet sells minutes as one-time purchases.

Heavy use of 10 hours a month fits a published plan on Narakeet, TTSMaker and Voicemaker. On Typecast it exceeds every published allowance, and only Business may buy extra credits.

What does each plan really count: characters, minutes or credits?

Each vendor counts a different unit, and on Fish Audio, Murf AI’s API and Voicemaker the voice you pick changes the rate. Narakeet counts minutes of finished audio in one-second increments, sold in packs of 30, 300, 1000, 2500 or 10,000 minutes. Murf AI’s Studio counts hours of Voice Generation Time, its API counts characters and Murf Dub counts credits at two per minute. TTSMaker counts characters, 300,000 a month on Lite, and Speechma’s free site counts characters per conversion, 2,000 standard or 1,000 for character voices. Typecast counts download credits spent only when you export, 3,000 lifetime on Free. Voicemaker counts one credit per character for default voices, with 200,000 a month on Starter, and Fish Audio counts credits, 250,000 on Plus.

Voicemaker’s expressive and high-resolution models cost four credits a character, and Fish Audio charges double for another team’s professional clone. Murf AI’s richer API model costs three times its base model per thousand characters, while Studio time is billed by audio length. Narakeet bills by output length whichever of its 900 voices you choose, and Speechma charges one character rate across its full voice catalog.

What does the free plan forbid or leave on my audio?

Most free plans keep you off paid work, though 6 of the 7 have one. Speechma and TTSMaker let you publish free audio commercially with no required credit.

Typecast requires a spoken or written credit naming Typecast and the characters used. Murf AI’s free account gives ten minutes of generation with no downloads, and free Murf Dub exports arrive watermarked. Fish Audio bars monetizing free output, Narakeet limits it to personal or evaluation use and Voicemaker to personal use.

Can I publish the audio commercially, and who owns it?

Commercial rights start on the first paid plan or earlier, and ownership is where the products differ. Speechma and TTSMaker allow commercial publishing from the free plan. Narakeet turns commercial with any paid top-up, Voicemaker from the $5 Starter plan, Typecast from Basic at $5 a month and Murf AI from its Creator plan.

Voicemaker lets you keep full copyright, and files made on a paid plan stay usable commercially after it ends, minus resale rights. Narakeet claims no copyright in your output and charges no royalty, and Murf AI’s commercial rights survive cancellation. Speechma gives a perpetual, royalty-free license, and TTSMaker a perpetual license that is usage rights, not copyright. Typecast gives a non-exclusive license, and after it lapses you keep using downloaded files but cannot export more.

What happens when I run out of characters or minutes?

Narakeet, TTSMaker and Voicemaker sell more usage past the allowance, the 3 of the 7 that do. TTSMaker sells one-off character add-ons to any subscriber. On Murf AI you upgrade or ask support to repurchase time, and on Speechma’s API you move to the next plan. Typecast lets only Business buy more credits on its web plans, while its API sells pay-as-you-go top-ups in fixed blocks.

Unused allowance carries over on 2 of the 7: Narakeet’s minutes never expire, and Murf AI carries unused time into the next cycle on an auto-renewing plan. Fish Audio and TTSMaker reset the allowance each cycle.

Can I clone my own voice, and will it train their models?

Fish Audio, Speechma and Voicemaker clone from a short sample, and they differ on how they check consent. Fish Audio needs no training step and checks professional clones with a spoken prompt or authorization documents. Speechma clones from a 5-to-30-second sample with no account and requires the speaker’s informed permission, banning celebrity clones. Voicemaker makes the speaker’s explicit consent your warranty. Typecast clones from Basic upward, and Murf AI has no self-serve cloning.

Fish Audio’s terms let it train on your content, with Enterprise zero data retention the one published way out. Narakeet claims no rights beyond processing your uploads and deletes source files after 30 days. Speechma keeps no recordings after processing, and Voicemaker trains on them only with separate research consent.

The tools that stand out

Fish Audio

Best for audiobooks and character voices: Story Studio assembles long-form, multi-speaker audio in a project editor.

  • Price: Free, then $15 a month
  • Free plan: yes
  • Who it’s for: Creators narrating videos, audiobooks and character voices from their own cloned voice, since cloning and Story Studio both run in the browser app.
  • Skip it if: You want subtitle files straight from the studio or help by live chat; Murf AI exports subtitle files and answers by live chat.

Fish Audio does text to speech, voice cloning and long-form production in one web app, with nothing to install. Story Studio is its project editor for audiobooks, dialogue and narration.

The free plan needs no card and gives 8,000 credits a month, capped at 500 characters per generation. Commercial use starts on Plus at $15 a month, which includes 200 minutes, so 2 hours of speech a month uses 120 of them. The pricing page has run a launch promotion on the first month, and paying yearly stacks a second discount on the headline price.

Plus, Pro and Max are bought by card on the pricing page, and only Enterprise volume pricing goes through sales.

What it does well

  • Story Studio puts a long, multi-speaker audiobook or drama in one project editor.
  • Plus at $15 a month covers 2 hours of speech a month with commercial use allowed.
  • A standing free plan gives 8,000 credits a month to try voices and cloning.
  • A connector plugs the account into Claude, Claude Code, Cursor, Codex and Windsurf.

Where it falls short

  • Support is email plus a Discord community, with no live chat and no published response times.
  • Subtitle files come from a developer recipe built on transcription timestamps, and the studio has no caption download.
  • The free plan caps each generation at 500 characters, and commercial use starts on Plus.

Murf AI

Best for dubbing video into other languages: Murf Dub opens an editor to correct the translation, transcript and voice settings before export, then exports translated subtitle files.

  • Price: From $29 a month
  • Who it’s for: Creators and e-learning teams producing voice-overs and dubbed video in many languages, who want to fix the translation before it ships.
  • Skip it if: You want to clone your own voice and pay by card, since cloning goes through sales; Fish Audio puts cloning in a web app you buy by card.

Murf AI sells a voice studio for scripts, Murf Dub for video and a separate speech API with its own free trial and pay-as-you-go plan. In Studio you import the video and stretch, trim or pad voice blocks on a timeline so each line lands on its clip. Scripts come in as Word, text or subtitle files and split into blocks, each regenerated on its own.

Studio, API and Dub plans are bought online with a card. Murf Dub limits how many projects run at once by plan and states no cap on file length or size.

Support runs live chat and email around the clock, where Fish Audio offers email and a Discord community. Business plans add a Windows voices installer for desktop apps.

What it does well

  • A dub editor lets you correct the translation, transcript and voice settings before export.
  • Studio and Murf Dub export subtitle files or burn subtitles into the video.
  • A timeline lets you stretch, trim or pad each voice block to land on its clip.
  • Live chat and email run around the clock, with priority support from the Creator plan.

Where it falls short

  • Voice cloning goes through sales.
  • Past your Studio allowance there is no top-up rate, so you upgrade or ask support to buy the plan again.

Narakeet

Best for buying minutes once, no subscription: Narakeet sells minutes as one-time packs, with no annual term and no recurring charge.

  • Price: Free, then $6 per 30 minutes
  • Free plan: yes
  • Who it’s for: Course makers and marketers turning slide decks and documents into narrated audio or video, who buy minutes when a project needs them.
  • Skip it if: You want to try subtitles or the API on a free account, since free accounts get neither; Speechma exports subtitle files on its free site.

Narakeet turns documents and slide decks into narrated audio or video in the browser, including on phones and tablets. Long documents run as projects, and there is an installable command-line client for batch work.

Minutes are paid once by card, PayPal or regional methods. Any paid top-up adds the API and the command-line client. Extra people join by emailing support and share the account’s credit balance, with no per-seat charge.

Free accounts cap uploads at 10 megabytes and a single transcription at 10 minutes, while commercial accounts allow 350 megabytes and 60 minutes per request. Dubbing produces an audio file to lay over your existing picture.

What it does well

  • Minutes are bought as one-time packs.
  • Paid accounts get subtitle files alongside the audio, included in the price.
  • A team shares the account’s credit balance, and support adds people by email.
  • Commercial accounts take uploads up to 350 megabytes and 60 minutes per request.

Where it falls short

  • Free accounts cannot get an API key or make subtitles.
  • Support is email only during British business hours, with no live chat or phone line.
  • Dubbing leaves the speaker’s mouth as it was, with no lip sync.

Speechma

Best for a permanently free voice studio: Its text to speech, dubbing and transcription stay free for good, with no account, no trial and no card.

  • Price: Free, then $9 a month
  • Free plan: yes
  • Who it’s for: Creators and small businesses who need commercial voice-over without an account or a budget.
  • Skip it if: You narrate books, since each generation takes 2,000 characters and long scripts must be split; Fish Audio’s Story Studio keeps long-form audio in one project editor.

Speechma runs text to speech, cloning, dubbing and transcription in any desktop or mobile browser, and it has a free Android app. The free tools need no account, and Speechma says the free price is permanent.

Dubbing takes a video up to 60 megabytes per dub, from 12 source languages into 9 target languages, free. Transcription exports a subtitle file with per-sentence timestamps. Lip movements stay untouched, and Speechma recommends dubbing for narration and talking-head footage.

The paid side is a separate API subscription with an instant key and 1,000+ voices, from $9 a month. Past the monthly quota no extra characters are sold, so you move up a plan.

What it does well

  • The website stays free for good and opens straight in the browser.
  • Dubbing handles videos up to 60 megabytes, from 12 source languages into 9 target languages, on the free site.
  • Transcription exports a subtitle file with per-sentence timestamps.
  • It runs in desktop and mobile browsers and in a free Android app.

Where it falls short

  • Each generation takes 2,000 characters, so a long script is split into parts and downloaded together in one archive.
  • Support is a single email inbox with no live chat, aiming to reply within a few business days.
  • There are no accounts, so a team gets no shared workspace.

Prices compared

Figure 1 · Entry price, a month Entry prices in US dollars start at $5 a month
  1. TK 01Typecast$5 a month
  2. TK 02Voicemaker$5 a month
  3. TK 03Speechma$9 a month
  4. TK 04Fish Audio$15 a month

Source: Each vendor’s pricing page, read between 26 October 2026 and 26 October 2026.

Typecast and Voicemaker open at $5 a month. On Voicemaker, Typecast and Murf AI the entry plan is where commercial rights begin, so for them the entry figure is the price of publishing.

“Starter $ 9 per month · ≈ ₹864 INR 1M Chars / month”

From Speechma’s own page, read on 26 October 2026

Which one to pick

  • If you need the smallest monthly bill with commercial rights: pick Voicemaker. Voicemaker grants commercial and distribution rights from its $5 Starter plan at $5 a month, and that plan also exports subtitle files.
  • If you need to dub your videos into other languages: pick Murf AI. Murf Dub returns your audio or video dubbed into the languages you pick, opens an editor to correct the translation before export and exports translated subtitle files.
  • If you need to narrate audiobooks in your own cloned voice: pick Fish Audio. Fish Audio clones a voice from a very short sample with no training step, and its Story Studio holds long, multi-speaker projects.
  • If you need a team to turn slide decks into narrated video: pick Narakeet. Narakeet turns slides into video with narration fitted to each scene, extra people share the account’s credit balance, and minutes are bought once.
  • If you need commercial voice-overs with no budget at all: pick Speechma. Speechma has a free plan that allows commercial publishing, covering monetized YouTube, ads, courses and client work, and its free audio carries no required credit.

How we compared

We compared 7 tools on price, free plan and 3 questions a buyer asks, reading each vendor’s pricing page and product pages on 26 October 2026. We did not sign up for or run any of them. The median, the cheapest and the dearest plan are taken over the 6 picks priced in US dollars a month, so their prices compare directly. In all we answered 40 questions about what they do.

Questions buyers ask

Which AI voice generator can I use for free and still publish commercially?

Speechma and TTSMaker both have free plans that allow commercial publishing, and neither requires a credit on the audio. Free plans are widespread, with 6 of the 7 offering one. Fish Audio, Narakeet and Voicemaker keep free output to personal or evaluation use, and Typecast requires a spoken or written credit on free downloads.

Do unused characters or credits roll over to next month?

Yes on Murf AI and Narakeet, the 2 of the 7 where unused allowance carries over. Narakeet’s minutes have no expiry, and Murf AI carries unused Voice Generation Time into the next cycle on an auto-renewing plan. Fish Audio and TTSMaker reset each cycle. Voicemaker gives none on Starter and Creator, one billing cycle on Pro monthly and up to three months on Teams and Business.

Can I clone my own voice, and do they train on it?

Fish Audio, Speechma and Voicemaker offer instant cloning, with Speechma working from a 5-to-30-second sample. Murf AI has no self-serve cloning on any paid Studio plan. Speechma processes your recordings only to make the result and keeps none afterward, and Voicemaker says it does not train its general voice models on them unless you give separate research consent. Fish Audio’s terms let it train on your content, and Enterprise zero data retention is the one published way out.

Which voice tools export captions or subtitles?

Murf AI, Speechma and Voicemaker export subtitles, the 3 of the 7 that do. Murf AI exports a subtitle file alongside the video or burns subtitles in, and Murf Dub exports translated ones. Speechma builds a subtitle file from transcription with per-sentence timestamps, and Voicemaker exports subtitle files from its Starter plan.

Can I produce a full audiobook with an AI voice?

Yes, on Fish Audio, Murf AI and Narakeet, the 3 of the 7 built for long scripts and chapters. Narakeet takes chapters as Word or ebook files and returns finished audio files, and Murf AI narrates with a voice per character in blocks you regenerate one at a time. Fish Audio’s Story Studio holds long, multi-speaker projects in one project editor. Murf AI and Narakeet state no retail store delivery spec, so check the store’s requirements before the final render.

Which AI voice tools have an API?

Murf AI, Speechma and Typecast sell a documented API on plans separate from the studio. Narakeet includes its API and a command-line client with any paid top-up, while free accounts get no key. Voicemaker’s API sits on a separate pay-as-you-go developer platform, and Murf AI’s covers text to speech, speech to text, translation and voice changing.

Sources

Top of text to speech and AI voice-over

The category page
Show the whole category

The Dubsheet digest

New voice tools, price changes and what we published, by email.

Privacy policy

More of our research

Methodology

How we price an hour of finished speech

Every AI voice tool counts something different. How Dubsheet turns credits, characters, seconds and minutes into one price for an hour of speech you may publish.