Cloning makes sense when the voice is part of the product: a founder’s podcast, a course instructor, a brand that has one recognizable narrator. If nobody knows whose voice it is, a good stock voice will beat a mediocre clone.

What audio does a clone need?

Quality beats quantity. Five minutes of clean audio outperforms thirty minutes with room echo.

Record in a small room with soft furnishings. A closet full of clothes beats an empty office — reflections are the number one killer, and you will not hear them until the clone does.

How do you record clone-ready audio?

1. Use a decent dynamic mic. A $100 dynamic mic in a quiet room outperforms a $500 condenser in a reverberant one. Proximity effect also helps vocals sound fuller.
2. Speak 15-20 cm from the capsule, slightly off-axis to reduce plosives.
3. Read continuous prose, not word lists. Natural sentence intonation is what the model learns.
4. Record 8 minutes, keep the best 4. Record extra and cut sections with coughs, hesitations, or chair squeaks.
5. Normalize to about -16 LUFS before upload. Not required, but it makes the clone level-consistent with stock voices.
6. Check the waveform for clipping before upload. Clipped audio teaches the model distortion.

What are the legal and ethical rules?

Cloning your own voice is straightforward. Cloning anyone else’s requires explicit written permission, and most platforms require a verification recording where the person reads a consent phrase.

Play.ht has a voice verification flow for this. Do not try to work around it. A cloned voice of a public figure used in synthetic media is the fastest route to a legal letter, and platform terms permit deletion of the asset and account.

Two practical rules for teams:

• Get consent in writing even from employees, and specify the allowed uses. A departing employee can revoke consent.
• Label synthetic audio. A short spoken intro or a description note that the narration is AI-generated. Listeners are forgiving about disclosed synthetic voice and hostile about undisclosed.

How does a clone compare to a stock voice?

Where a clone wins: consistent brand identity across 100 episodes, pronunciation of unusual names, and emotional delivery in first-person writing.

Where a stock voice wins: cost, speed, and emotion range on scripted marketing copy. Clones tend to flatten excitement and struggle with questions — the model learned your declarative narration.

A hybrid works better than either pure approach: the clone narrates long-form explainers where consistency matters, and stock voices handle promos and ads where energy does.

What goes wrong after training?

Sibilance. Clones amplify “s” sounds, especially if the source had a bright mic. A de-esser in any editor fixes it, and you should apply it on every render.

Sentence-start drift. Long sentences starting with the same word sound mechanical. Vary your script openings.

Numeric reading. Clones inherit your number pronunciation habits from the sample. If the source never said “2026”, expect odd readings — fix them in the dictionary.

Emotional range collapse. Ask a clone to sound surprised and you get louder, not surprised. Write exclamation-free copy for clones.

When should you not clone?

If you publish fewer than about 10 audio pieces a year, skip it. The setup time and per-render review overhead will exceed the benefit, and a well-chosen stock voice is indistinguishable to most listeners on a five-minute explainer.

Same for multilingual content. A clone of an English speaker narrating Spanish produces an accent that sounds wrong to native listeners. Use native-language stock voices instead.

Key Takeaways

• A usable Play.ht clone needs 2-5 minutes of clean, continuous speech in a low-reflection room.
• Recording quality matters far more than sample length; a closet beats an office.
• Consent must be in writing for any voice other than your own, and Play.ht enforces verification.
• Clones flatten emotional range — write calmer copy and use stock voices for high-energy spots.
• Under about 10 audio pieces a year, a stock voice is the better economics.

Clone types and tier limits are listed on the Play.ht page at iAIFeed.

Audio versions of posts are worth doing for one reason — a second surface for the same content, on platforms where competition is thinner than search.

Which voice should you pick?

The default recommendation is a mid-range voice with a neutral accent and moderate pace. Enthusiastic voices with wide pitch range sound uncanny on long-form text; flat voices put listeners to sleep by paragraph four.

Three rules that hold up:

• Match the content register. Technical explainers suit a calm, slightly slower voice. Marketing copy tolerates more energy.
• Preview the same 100 words across five voices. Do not judge voices on the app’s sample sentences; they are crafted to flatter every option.
• Skip cloning until you have clean source audio. A clone trained on a noisy podcast sounds worse than a good stock voice.

How do you fix pronunciation?

This is where most auto-audio fails. Product names, acronyms, and version strings get read wrong, consistently.

Play.ht accepts SSML tags, and the practical set is small. Wrap a word in a phoneme tag to force its pronunciation, use the characters mode for acronyms you want spelled out letter by letter, and insert explicit pauses where the script needs a beat.

Build an account-level pronunciation dictionary for anything you say repeatedly: brand names, competitor names, technical terms. Add each entry the first time it comes out wrong. After five posts the dictionary covers most of your vocabulary.

Numbers are their own problem. “1,200” and “2026” read fine; part numbers and version strings do not. Write them out in the script exactly as you want them spoken.

How long should the audio be?

Not the whole article. A 3,000-word post becomes 20 minutes of audio, and completion drops off steeply past 12 minutes.

Two working formats:

For full-article audio, open with a spoken promise of what the listener will be able to do by the end. Written posts can be skimmed; audio cannot.

What does Play.ht cost?

Pricing is credit-based, with credits mapping roughly to characters of text. The free tier covers a couple of short posts. Paid plans start around $39 a month for a few hundred thousand characters and climb into enterprise tiers above that.

Do the math on volume before committing. A 1,500-word post is roughly 9,000 characters, so 20 posts a month is about 180,000 characters. That fits a mid-tier plan — until you account for re-renders. Budget two to three renders per post while you are still fixing pronunciation, which is closer to 400,000 characters.

What are the honest limitations?

Emphasis. The voice will not stress the right word in a sentence like “it is not the speed that matters, it is the accuracy.” You can force it with SSML per sentence, but you cannot batch that across a whole post.

Long-form fatigue. Even good voices get repetitive after ten minutes. Varying pace between sections helps, and so does a short spoken transition into each new section.

Podcast-ready metadata. ID3 tagging and chapter markers work but need setup. If you plan to distribute as a podcast, do it up front — retrofitting across 50 episodes is miserable.

Should every post get audio?

No. Tutorials with code blocks, comparison tables, and click-by-click instructions do not survive audio. A listener cannot follow “open the third dropdown in the panel.”

Save audio for concept explainers, opinion pieces, and guides with a linear narrative — usually 30-40% of a content calendar. Enough volume to matter, small enough to keep quality high.

Key Takeaways

• Play.ht renders a 1,500-word post to MP3 in about a minute with no recording setup.
• Choose a neutral mid-range voice and test it on your own 100 words, not the built-in samples.
• Build a pronunciation dictionary early for brand and technical terms; add an entry every time a word comes out wrong.
• Keep audio under about 12 minutes; long full-article reads lose listeners.
• Only convert narrative content — code and UI tutorials fail in audio.

Voice options and plan limits are on the Play.ht page at iAIFeed.