# HeyGen Voice Cloning: Instant Clone vs Professional Voice Clone > HeyGen voice cloning explained: free instant clones from 60 seconds of audio, the per-voice Professional Voice Clone add-on, plan limits and consent gaps. _Source: https://professionalstoolkit.com/articles/heygen-voice-cloning — The Professional's Toolkit · updated October 2026_ --- HeyGen gives you two ways to clone a voice, and they solve different problems. The **Instant Voice clone** takes up to 60 seconds of audio from any device, is free to create in the app and is ready in minutes.[1] The **Professional Voice Clone** (PVC) is a paid add-on trained on at least 20 minutes of recordings, with quality guaranteed in 11 languages. Instant clones count against a per-plan limit: one on Free, up to 40 on Creator, up to 70 on Pro, unlimited on Enterprise. Our verdict is simple. Start with an instant clone, and pay for PVC only when a recurring narrator voice has to hold up across long, polished videos. **60 seconds** — Maximum audio HeyGen takes for a free instant voice clone in the app. - **Up to 70** — Voice clones on the Pro plan (Creator: up to 40) - **20+ min** — Recordings a Professional Voice Clone needs - **11** — Languages with guaranteed Professional Voice Clone quality ## Instant clone or Professional Voice Clone: pick by the job The two clone types differ in input, cost and control, and an instant clone cannot be upgraded later. A PVC is built from scratch from new recordings, so deciding early saves a second recording session. The grid below sums up what HeyGen's help center and developer docs say about each. | | Instant Voice clone | Professional Voice Clone | |---|---|---| | Audio needed | Up to 60 seconds, one recording | 20+ minutes; 30 minutes to 2 hours works best | | Ready in | Minutes | Training usually 20 to 30 minutes | | Cost | Free to create in the app | Paid plan plus the PVC add-on, per voice | | Retraining | Not possible | Five pooled trainings per slot each month | | Languages | API cloning and speech: about 40 base languages | 11 with guaranteed quality | | Fine controls | Similarity, engine choice | Speed, Volume, Pitch, Inflection | HeyGen's help center lists the Professional Voice Clone add-on at $29 a month or $288 a year per voice, and it needs a paid plan underneath.[2] Credits, not the plan or add-on price, decide how much finished video that voice ends up in, which our [HeyGen pricing guide](/articles/heygen-pricing) works through. Twelve monthly payments come to $348, so yearly billing saves $60 per voice by our arithmetic. Yearly already comes out cheaper if you keep the voice 10 months or more, since 10 monthly payments reach $290. The instant route suits founders, marketers and trainers who want their own voice on a talking avatar this week. The professional route suits a brand voice that narrates dozens of modules, where small drifts in tone across videos would be noticed. > ⚠ **Instant clones cannot be upgraded:** A HeyGen instant clone cannot be turned into a Professional Voice Clone; the professional version is trained from new recordings of 20+ minutes. ## Clone limits on each plan, and how to free a slot HeyGen's pricing page lists Creator at $29 a month billed monthly or $24 a month billed yearly ($288 a year), and voice cloning is one of the features it adds over Free. Credits, not the plan price, decide how much avatar video those voices actually narrate, as the [pricing guide](/articles/heygen-pricing) explains. The clone allowance then scales with the plan: Free gets one voice clone, Creator up to 40, Pro up to 70, Business the same maximum as Pro, and Enterprise an unlimited number. **Extra instant-clone slots cannot be bought.**[3] When you reach the cap, you delete voices you no longer use; HeyGen's help says the freed slots show up after you log out and wait a few hours. A voice that a template still uses cannot be deleted, so check templates first. The web app and the API share one clone pool, and an API call that hits the cap returns a limit error asking you to delete unused clones or contact support. One exception softens the cap: on paid plans, voices created during avatar creation do not count against the allowance. The Free plan's single clone, like everything else made on Free, is for personal, non-commercial use only. > 💡 **Clone from your cleanest take:** Record in a quiet room in one continuous take; HeyGen's help says more and cleaner audio is the main fix for a clone that sounds off. ## Recording an instant clone that sounds like you The workflow lives in AI Studio now, because HeyGen removed the old homepage Voice tab; the same New Voice menu also offers Generate a New Voice, which designs a voice from a text description when you would rather not record at all. For a clone of your own voice, the steps are short: 1. Open AI Studio, choose New Voice, then Instant Voice Cloning. 2. Record or upload up to 60 seconds of clean speech in one take. 3. Pick the clone in the voice menu and adjust its settings. 4. If the result sounds off, re-record through Improve Voice. Quality comes almost entirely from the sample. HeyGen advises re-cloning with more and cleaner audio, and Improve Voice needs at least 30 seconds of new recording. For cloned voices, the help center suggests raising the Similarity slider from its default of 75 toward 95. It also points English speakers to the Fish engine for the closest clone. If the clone sounds flat or monotone, add a high-energy recording as a Custom Emotion. A Digital Twin brings its own voice. **Every video avatar includes voice cloning**, built automatically from the training footage, and that twin needs an on-camera consent clip from the person depicted. For Avatar V, HeyGen still strongly recommends a standalone voice clone. Its help center names the voice as one of three inputs that set Avatar V quality. Deleting the avatar later does not delete the cloned voice, which is removed separately. ## Professional Voice Clone: uploads, training and the consent gap The PVC flow is upload only. You cannot record inside the app, so prepare files first: up to 30 files in the app (10 through the API), 1.5 GB in total, as MP3, WAV or M4A audio, or MP4 and MOV video. Every file must be the same speaker. Training usually finishes in 20 to 30 minutes, and each purchased slot carries five trainings per monthly billing period, pooled across your slots. Retraining counts against that pool, while failed trainings are free. Languages are the real filter. HeyGen guarantees PVC quality only in English, Spanish, Portuguese, French, German, Italian, Russian, Mandarin, Cantonese, Japanese and Korean. Anything else may work, without that promise. If your audience speaks Hindi or Arabic, test an instant clone first. **The PVC flow has no consent step.** HeyGen's help says so plainly, while a Digital Twin cannot render without the depicted person's on-camera consent. That puts the burden on you: get written permission before uploading anyone's voice, including a former employee's or a hired narrator's. Synthesia, by contrast, says it requires verbal consent during the cloning process and lets you clone only your own voice. Data handling also applies here. HeyGen's biometric notice says voiceprint data may be collected from avatar footage and used for training only if you consented. Its Terms grant a licence to train on your content; its Trust page gives non-Enterprise users an opt-out, and Enterprise data is excluded. Our [safety review](/articles/is-heygen-safe) covers those layers. > ⚠ **No consent check on PVC uploads:** HeyGen's Professional Voice Clone flow is upload only with no consent step, so get written permission from the speaker before you upload their recordings. ## Directing the clone: engines, Voice Director and Voice Mirroring A clone is only half the result. The engine that speaks it matters as much. Studio now names the active engine instead of showing Auto. HeyGen's help lists four engines. ElevenLabs runs in V2 and V3 versions. Panda powers Voice Director and Voice Mirroring. Starfish targets Chinese, Japanese and Korean, while Fish, built on fish.audio, handles accents and locales. Similarity, stability and style sliders appear only on ElevenLabs Multilingual v2 or Fish; volume and speed work everywhere. Voice Director shapes delivery with presets such as Excited, Calm, Serious or Sarcastic, opened by typing a slash in the script, plus your own text instructions. A direction can apply to one scene or to the voice across every scene. It requires Panda, and it works with HeyGen clones and imported ElevenLabs voices. Voice Mirroring goes further: you record or upload an MP3 or WAV up to 100 MB, HeyGen transcribes it, and the avatar copies your pacing, accent and emotion in the chosen voice. A Mirroring recording can itself become a new clone. Voice previews and text-to-speech regenerations cost no credits, so tune freely before rendering. Only the avatar render draws on plan credits. If you already own a strong clone elsewhere, HeyGen imports ElevenLabs and LMNT voices with your own API key. For accents an instant clone misses, HeyGen's help even suggests building a professional clone in a paid ElevenLabs account. ElevenLabs lists Professional Voice Cloning from its Creator plan at $22 a month; its $6 Starter plan has instant cloning only. > 💡 **Tune voices before rendering:** Voice previews and text-to-speech regenerations cost no credits in HeyGen, so adjust Similarity, engine and Voice Director before you render the avatar. ## Cloning through the API: a separate wallet Developers clone voices through the v3 endpoints, and that usage bills from a prepaid API wallet, not from plan credits. The wallet starts at $5, and purchased API credits expire after 12 months. Instant clones are free to create while the API preview lasts and use only the first 3 minutes of a recording. Professional voices need one purchased slot per voice on the API Usage page, and synthesis bills at 0.6 API credits per generated minute. HeyGen Voice text-to-speech runs $30 per 1M characters, and Voice Turbo $15 per 1M characters for public and instant voices. The Turbo Free model costs nothing until November 30, 2026, but it rejects professional clones. Speech-to-speech, the API form of Voice Mirroring, bills $0.12 a minute and converts only into instant voices for now. Our [HeyGen API guide](/articles/heygen-api) covers the wallet and rate limits in full. ## FAQ **Can I clone someone else's voice in HeyGen?** HeyGen's Professional Voice Clone upload has no consent step, so the platform does not check; the legal and ethical burden is yours. Get the speaker's written permission first, and remember a Digital Twin of another person needs that person's own on-camera consent. **Does deleting my HeyGen avatar delete my cloned voice?** No. HeyGen's developer docs say the cloned voice is its own resource; removing an avatar leaves the voice, footage and consent video in place until each is deleted separately. **Can I use my HeyGen voice clone for commercial videos on the Free plan?** No. Output from HeyGen's Free plan is for personal, non-commercial use only; commercial use starts on Creator, which HeyGen's pricing page lists at $29 a month billed monthly or $24 a month billed yearly. Credits, not the plan price, decide how much video you get. **Can I bring an ElevenLabs voice into HeyGen?** Yes. HeyGen imports ElevenLabs and LMNT voices with your own API key; the key needs Text to Speech, Voices, Voice Generation, Models, History and User permissions, or Restrict Key turned off. ## The bottom line HeyGen's voice cloning is generous at the instant tier and strict about money at the professional one. An instant clone covers most talking-avatar work, costs nothing to create and sits inside a plan allowance that runs from one voice on Free to up to 70 on Pro. Pay for a Professional Voice Clone only for a long-lived narrator in one of its 11 guaranteed languages, budget the add-on per voice, and secure written consent yourself, because HeyGen's upload flow will not ask for it. ## Sources [1] HeyGen Voice 1: Text to Speech, Speech to Speech, Voice Cloning Skip to main content - Platform Use cases Developers Resources Enterprise Re — https://www.heygen.com/heygen-voice (2026-10) [2] HeyGen Help: how to create a great professional voice clone pvc — https://help.heygen.com/en/articles/17589678-how-to-create-a-great-professional-voice-clone-pvc (2026-10) [3] HeyGen Help: how to get started with voices — https://help.heygen.com/en/articles/9834925-how-to-get-started-with-voices (2026-10) [4] HeyGen pricing page — https://www.heygen.com/pricing [5] HeyGen Help Center — https://help.heygen.com/en/ [6] HeyGen Terms of Service — https://www.heygen.com/terms [7] App Store (US), HeyGen listing — https://apps.apple.com/us/app/ai-avatar-generator-heygen/id6711356409