Up to 5,000 characters per generation. Punctuation shapes the pacing of the read.
0 / 5,000 characters
Standard gives the most natural delivery. Turbo is faster and costs fewer tokens.
Choose one of 39 languages, or let the text decide with auto-detect.
Public library voices. Ignored when reference audio is supplied.
Clone a voice from a single-speaker recording up to 180 seconds (MP3, WAV, M4A, AAC, or OGG, up to 25 MB). When provided, the library voice is ignored.
Sign in to spend tokens
Your generated audio will appear here
Prizmad's HeyGen Voice turns a written script into natural speech in 39 languages, or clones a voice from a short reference recording. Standard costs 30 tokens per generation and Turbo costs 4, both capped at 5,000 characters and returned as a 44.1 kHz WAV file.
HeyGen Voice is Prizmad's studio-grade voiceover tool. Paste up to 5,000 characters of copy, pick one of 24 public library voices, choose the language, and generate a ready-to-use WAV file in seconds. It is built for ad scripts — hooks, benefit stacks, and calls to action — but works for any narration you would put behind a video.
The second mode is instant voice cloning: upload a clean single-speaker recording of up to three minutes and the tool returns speech in that voice instead of a library voice. That is how brands keep a consistent narrator across dozens of variants without booking a studio session.
Two quality tiers keep cost aligned with the job. Standard is the most natural delivery and suits hero ads and final takes. Turbo is faster and dramatically cheaper, which makes it the right choice for hook testing, internal drafts, and high-volume localization runs.
On Prizmad the output lands in your asset library next to your scripts, videos, avatars, captions, and music, drawing from the same token balance. The file is a 44.1 kHz WAV, so it drops straight into an editor or the video montage tool without a conversion step.
Read benefit-led ad scripts over product footage, slideshow creatives, and UGC-style clips.
Clone your spokesperson once and generate every variant — new hooks, new offers, new markets — in the same recognizable voice.
Ship the same script in Spanish, German, Japanese, Arabic, or any of the other supported languages without re-recording.
Generate ten opening lines on Turbo for a fraction of a Standard take, then re-record the winner on Standard before launch.
Narrate how-it-works videos and landing-page explainers directly from documentation or a product brief.
Enough for a 60–90 second ad read or a long product narration in a single generation.
Arabic, Belarusian, Bulgarian, Catalan, Czech, Danish, Dutch, English, Filipino, Finnish, French, German, Greek, Hebrew, Hindi, Croatian, Hungarian, Indonesian, Italian, Japanese, Korean, Macedonian, Malay, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Serbian, Spanish, Swedish, Tamil, Thai, Turkish, Ukrainian, Vietnamese, and Chinese — or let auto-detect read the language from your text.
A curated public voice library covering warm narrators, bright commercial reads, and calm premium tones. Pick a name and generate; no voice training required.
Upload a single-speaker recording of up to three minutes and generate speech in that voice. No training queue and no separate model to manage.
Standard gives the most natural delivery for hero creative; Turbo trades a little polish for speed and a much lower token price.
Every generation returns a broadcast-ready WAV file plus its duration, so you can plan cuts and captions without opening a separate audio editor.
Open the HeyGen Voice workspace in AI Studio (or write the script first with the free Script Writer tool).
Paste your text — up to 5,000 characters. Commas and periods shape the pacing, so punctuate the way you want it read.
Choose the language, or leave auto-detect on and let the text decide.
Pick a quality tier: Standard for the final take, Turbo for drafts and volume testing.
Either select a library voice, or upload a reference recording to clone a voice instantly.
Click Generate. The WAV file appears in your asset library with its duration, ready to attach to a video or a montage.
Standard generations cost 30 tokens and accept up to 5,000 characters — a full ad read in one go, with the most natural delivery the tool offers.
Turbo generations cost 4 tokens for the same 5,000-character limit. That is the cheapest way to test hooks, draft scripts, and produce localized variants at volume.
HeyGen Voice draws from the same token balance as your script, video, avatar, caption, and music tools. If a campaign drains it, a one-off top-up is available from your billing page.
No — a generation costs tokens from your balance: 30 for Standard and 4 for Turbo. The Script Writer tool is free, so going from a product description to a finished voiceover can cost as little as 4 tokens.
Up to 5,000 characters per generation, which is roughly 60–90 seconds of speech. Longer narration needs to be generated in sections of up to 5,000 characters each.
39 languages are supported, including English, Spanish, German, French, Portuguese, Italian, Dutch, Polish, Russian, Ukrainian, Turkish, Arabic, Hebrew, Hindi, Tamil, Japanese, Korean, Chinese, Vietnamese, and Thai. Auto-detect reads the language from your text when you do not pick one.
Yes. Upload a clean recording of a single speaker of up to three minutes and the tool generates the script in that voice. The reference clip replaces the selected library voice for that generation.
Standard is the highest-fidelity delivery and the right choice for final creative. Turbo is faster and much cheaper, which suits drafts, hook testing, and high-volume localization. Both accept the same 5,000 characters and return the same 44.1 kHz WAV format.
A 44.1 kHz WAV file together with its duration in seconds. It is stored in your asset library and can be attached to a video or a montage directly.
Yes. Voiceovers you generate on Prizmad are yours to use in paid ads, organic videos, product pages, and other marketing content. Only upload reference recordings you have the rights to use.
The tokens for that generation are returned to your balance automatically, so a provider hiccup never costs you credits.
Published 2026-11-01 · Last updated 2026-11-01