SteosVoice
SteosVoice is a neural text-to-speech and voice-cloning platform from AGI lab Mind Simulation, offering 800+ voices, 44.1 kHz WAV output and an API, with a free Telegram tier and paid plans starting at USD 2 per month.
What is SteosVoice?
SteosVoice, which still answers on its former CyberVoice domain, is a neural speech synthesis service built by Mind Simulation, an independent AGI laboratory credited in every page footer. Three capabilities sit at its core: text-to-speech with voice imitation, re-voicing of an existing audio file, and an API that exposes both to developers.
The scale comes from the site's own counters: more than 800 voices and growing, 1.6 million users, and roughly 75 hours of audio generated every day. The parent lab's CEO page puts the figure higher still, at over two million users across forty countries, and describes STEOS as profitable and entirely bootstrapped. Audio is delivered as 44.1 kHz WAV, and paying accounts can switch to MP3.
There are two front doors. The free one is a Telegram bot, @steosvoice_bot, which grants a permanent daily allowance of 1,000 characters without a card. The other is the web platform, entered through a single STEOS ID account. Everything is metered in tokens: 100 characters of synthesis costs one token, one second of re-voicing costs one, and API usage is priced separately.
What sets the product apart is its marketplace. Voice actors and owners of distinctive voices license digital versions of their voice and collect 20% royalties automatically, with no cap on how many variants one author publishes. The site names the talent involved, among them Andrey Zaitsev, the Russian dubbing voice of Anakin Skywalker and Will Turner, and American narrator Gary Ladka, and lists customers ranging from YouTube channels to game studios such as Norllina Studio and educational publisher BRAINY KIDS GAMES.
The origins are unusual: a viral living Geralt demo built around The Witcher 3 needed a voice layer, and that layer became the product. Coverage of the resulting AI-voiced mod, from Kotaku to Vice, is reprinted across the site. Legally the picture is split: the terms name MSL LAB LLC in Yerevan, Armenia as the administration, while every payment is routed through a Russian company.
What it does
- Turn written text into natural-sounding speech using a library of more than 800 voices
- Re-voice an existing audio recording with a different synthetic voice
- Clone your own voice from 15 to 20 seconds of recorded speech
- Dub YouTube videos, game mods and indie titles into another language or character voice
- Produce podcasts, audiobooks and audio versions of written articles
- Voice stream donations, patron greetings, IVR menus and tutorial videos
- Call the speech synthesis API to embed a voice inside your own product
When to use SteosVoice / When not to
A quick filter to help you decide if SteosVoice is the right fit.
When to use SteosVoice
- YouTube creators and video localisers who need voice-overs in a language they do not speak themselves
- Game modders and indie studios voicing quests, characters and re-dubbed dialogue
- Podcasters, audiobook producers and publishers turning written material into narrated audio
- Voice actors and owners of distinctive voices who want to license a digital twin and collect royalties
- Small brands and media teams needing a recognisable brand voice, IVR or article voice-over on a tight budget
When not to use SteosVoice
- Organisations with GDPR obligations: no DPA, no subprocessor list and no Article 27 representative is published
- Anyone planning commercial output from the free tier, which the plans page explicitly forbids
- Donation platforms wanting to voice donations, excluded from standard plans without a separate written agreement
- Buyers who need unused credit to roll over: tokens expire on schedule and are never refunded
- Users expecting a native iOS or Android app, or voice selection inside Telegram Web, Telegram X or Telegram Lite
How to use SteosVoice
A typical end-to-end flow, from setup to results.
- Open @steosvoice_bot in Telegram and send /start to use the free tier without registering
- Or follow Get Started to the web platform and create a STEOS ID account at id.steos.io
- Confirm your email address; only one account per person is permitted
- Paste or type the text you want spoken, having checked that you hold the rights to it
- Pick a voice from the library and check its personal or commercial usage marking
- Type + before a stressed vowel to fix pronunciation, and use full stops and commas to place pauses
- On a paid tariff, adjust narration speed and pitch and choose MP3 or WAV output
- Synthesise the text, download the file and keep an eye on the on-screen token balance
- Buy more characters inside the bot with the /symbols command when the daily allowance runs out
- To publish your own voice, open the my voices block, choose provide a voice, submit the form and wait for review
Pros & Cons
Pros
- Very low entry price: USD 2 a month buys roughly 1,222 minutes of speech, commercial use included
- Permanent free tier through Telegram, with no card and no time limit
- More than 800 voices, including named professional actors who are paid royalties
- 44.1 kHz WAV output rather than a compressed preview file
- A public API, plus a grant offering selected startups twelve months of free access
- Unusually explicit marketplace ethics: voice authors keep their rights and take 20% royalties
- Practical prosody controls for stress, pauses, narration speed and pitch
Cons
- No GDPR framework at all: the privacy policy runs under Russian law and the terms under Armenian law
- No Data Processing Agreement, no subprocessor list, no training opt-out and no stated minimum age
- Two different legal entities appear on the site, one Armenian and one Russian, the latter handling all payments
- Tokens expire at the end of the period and are never refunded, even when the account is deleted
- No iOS or Android application, and voice selection fails outside official Telegram clients
- Platform pages render entirely in JavaScript, and the console subdomain still cited in the FAQ no longer responds
- Published audience figures contradict each other and the footer contact link returns a 404
Pricing & Plans
A permanent free plan is available through the Telegram bot, limited to 1,000 characters per day and expressly barred from commercial use. The lowest paid entry point is USD 2.00 per month, rated by the plans page at approximately 1,222 minutes of speech and including download of all generated files and commercial use. Higher tiers are priced at USD 6.00 and USD 10.00 per month, and high-volume customers are invited to negotiate an individual plan.
- no charge
- 1
- 000 characters per day
- 800+ voices
- commercial use not permitted
- approximately 1
- 222 minutes of speech
- voice over text
- all files downloadable
- commercial use
- approximately 3
- 833 minutes of speech
- same inclusions
- approximately 6
- 650 minutes of speech
- same inclusions
- priced on request for high-volume commercial users
- includes brand voice creation
- twelve months free for selected startups and creators
- with 45
- 000 to 100
- 000 tokens
Data, GDPR & hosting
A consolidated view of how SteosVoice handles your data.
GDPR overview
There is no GDPR implementation to report. Neither the privacy policy nor the terms mention the GDPR, a data protection officer, a Data Processing Agreement, standard contractual clauses or an Article 27 EU representative. The privacy policy, dated 17 February 2023, states plainly that collection, processing, storage and disclosure are governed by the legislation of the Russian Federation, and relies on Federal Law 152-FZ to argue that separate consent and notification of the supervisory authority are not required. The terms of service, dated 29 October 2024, place content and disputes under the law of the Republic of Armenia. Users are nevertheless granted access, rectification, blocking and deletion rights through their account, plus a written request route answered within thirty days, which must carry identity-document details and a signature.
Who owns the data?
The terms split ownership three ways. The synthesised voices themselves belong to SteosVoice or to the users who supplied the source recordings, and speech generated by the site is described as SteosVoice property. The WAV files you download are yours: you retain ownership of the audio, together with a non-exclusive, limited licence over the embedded voice technology, provided you respect the terms and your payment obligations. In exchange you grant SteosVoice a free, worldwide, non-exclusive right to use content you post, in order to run the website, conduct its own internal research and development, improve its products, troubleshoot and perform general data analysis. Archived copies of that content may survive your deletion.
Reuse rights
Downloaded WAV files can be reused without asking further permission, for personal or commercial purposes, provided you are on a paid plan and comply with the terms. Free-tier audio is barred from commercial use, because royalties are shared with the voice authors. Individual voices also carry markings stating whether they are cleared for personal or commercial use, and ignoring those markings can lead to suspension or termination of the account. Two carve-outs remain: commercial rights under standard plans do not extend to integration with donation platforms, which requires a separate written agreement and a dedicated API, and SteosVoice reserves the right to remove a voice that breaches the rules.
Data retention & training
Hosting summary
No hosting country or region is named anywhere on the site. The privacy policy says only that traffic runs over standard SSL connections and that the service uses reputable standard hosting providers, without identifying them. Audio generated through the Telegram bot is the one documented exception: those files sit on Telegram's servers and fall under Telegram's own user agreement and privacy policy. The applicable jurisdictions are clearer than the infrastructure. The privacy policy of 17 February 2023 states that collection, processing, storage, disclosure and protection are governed by the legislation of the Russian Federation, and that personal data may be provided to state bodies under Russian procedure. The terms of 29 October 2024 place content, intellectual property and disputes under the law of the Republic of Armenia, except disputes with Russian entities, heard where the Russian payment company is located. One technical observation, offered as context rather than as a vendor statement: the domain's A record resolves to an address geolocated in Moscow.
Things to keep in mind
Risks and trade-offs to weigh before adopting SteosVoice.
- Voice imitation invites impersonation: the platform can remove a voice if you cannot prove it is yours or if it belongs to a public figure, but the damage may already be done
- Personal data is governed by Russian and Armenian law, with no GDPR framework and no named data protection officer to appeal to
- Voice authors hand over passport details, date and place of birth and bank card data in order to receive royalties
- The privacy policy reserves the right to share personal information with a trusted third party that combines it with data of its own
- Your name or company name and avatar are visible to every registered user, and voice authors are published under their full name
- Content you post is licensed to the operator for internal research, product improvement and data analysis, with archived copies possibly surviving deletion
- Leaning on synthetic narration can quietly erode the craft it replaces, and audiences increasingly expect disclosure when a voice is not human
Setup & Integrations
Technical difficulty
Very low for everyday use. The Telegram route needs no account at all: open @steosvoice_bot, send /start, pick a voice and paste your text. The web platform asks only for a single STEOS ID account and email confirmation. Two small learning curves follow: typing + before a stressed vowel to fix pronunciation, and using punctuation to place pauses, since pause length cannot be set. Speed, pitch and MP3 output unlock on a paid tariff. API integration is the one developer-level step, and it is billed separately in tokens.
Deployment
Integrations
Supported languages
Behind SteosVoice
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Is there a free version of SteosVoice?
How much does the cheapest paid plan cost?
How many voices are available?
Can I clone my own voice?
Does SteosVoice offer an API?
Is SteosVoice GDPR compliant?
How long are generated files kept?
Are unused tokens refunded or carried over?
Is there a mobile app?
Should you pick SteosVoice?
SteosVoice is one of the cheapest credible routes to broadcast-grade synthetic speech. Two dollars a month buys roughly 1,222 minutes of voice-over with commercial rights attached, the library runs past 800 voices, output arrives as 44.1 kHz WAV rather than a compressed sample, and a permanent Telegram tier lets anyone judge the quality before paying anything. The marketplace behind it is unusually explicit about ethics: voice actors license digital twins of their own voice, keep their rights and collect 20% royalties automatically, a stance the site backs by naming the professionals involved. For YouTube creators, game modders, indie studios, podcasters and small brands, that combination is hard to match at the price.
The reservations are serious and mostly legal. Nothing on the site mentions the GDPR: the privacy policy runs under Russian law and Federal Law 152-FZ, while the terms place content and disputes under Armenian law. No Data Processing Agreement, subprocessor list, training opt-out or minimum age is published, and the policy openly reserves the right to pass personal data to a trusted third party that combines it with its own. Two legal entities coexist on the site, MSL LAB LLC in Yerevan as the administration and a Krasnodar company for every payment, and the home page code labels the Russian one as old. Add tokens that expire without refund, a dead console subdomain still cited in the FAQ, a contact link that returns a 404, and audience figures that contradict each other.
The verdict: excellent value for creative and entertainment work where the audio is the deliverable, and a poor fit for any organisation that must document where personal data goes.
- Choosing a selection results in a full page refresh.
- Opens in a new window.