How to Migrate From Play.ht to ElevenLabs
Play.ht was an early mover in the AI voice space, and for a long time it was the go-to choice for podcasters who wanted a decent voice clone without a steep technical bar. The reasons people start looking elsewhere tend to cluster around two things: the emotional range of generated voices sounds flat on anything other than straightforward narration, and voice clones built from shorter audio samples often drift noticeably from the original speaker.
ElevenLabs has built a reputation for handling both of those problems better than most alternatives. That doesn't mean the switch is smooth, there are real workflow differences and at least one pricing gotcha that can catch people off guard. This guide covers what actually changes and how to move your existing projects across.
What's actually different
Play.ht and ElevenLabs both offer text-to-speech and voice cloning, but they've optimised for different things. Play.ht has a cleaner interface for non-technical users, a WordPress plugin for inline audio, and a broader set of voices in its standard catalogue. ElevenLabs has put most of its engineering effort into voice quality, emotional expressiveness, and the underlying models.
| Feature | Play.ht | ElevenLabs |
|---|---|---|
| Voice catalogue | 900+ standard voices | 900+ including community voices |
| Emotion/style control | Style selector (limited) | Stability + Similarity sliders |
| Voice cloning | Yes, ~30 sec minimum | Yes, ~1 min recommended |
| Clone quality at short samples | Variable | Generally stronger |
| WordPress plugin | Yes | No native plugin |
| API quality | Functional | More widely adopted |
The emotion control difference is meaningful in practice. Play.ht offers a style selector with options like "narrative," "customer service," or "angry." ElevenLabs doesn't use named styles; instead, the Stability and Similarity sliders shape how the model interprets the text itself. Lower Stability means the model reads more meaning into punctuation and context, which produces more natural emotional variance. Higher Stability produces consistent, controlled output for narration.
For voice cloning, ElevenLabs' Instant Voice Clone (IVC) and Professional Voice Clone (PVC) are two separate products. IVC works from as little as one minute of audio and is good enough for most use cases. PVC requires submitting audio for manual training and is priced separately, it's aimed at people who need a commercially licensed clone of a specific voice for large-scale distribution.
Mapping your existing workflow
Play.ht organises projects as documents with embedded audio. When you edit text, you can regenerate just the changed section without regenerating the whole file. ElevenLabs doesn't have a document-style editor; you work at the paragraph or section level and assemble the full audio externally.
If you use Play.ht's WordPress plugin to auto-generate audio for blog posts, you'll need to replace that with the ElevenLabs API or a third-party connector. There are Zapier and Make automations that can trigger ElevenLabs generation when a new post is published, but it's not as turnkey as the native Play.ht plugin.
Your voice clones from Play.ht don't transfer. You'll need to re-upload the source audio to ElevenLabs and regenerate the clone. If you still have the original recordings you used to train the Play.ht clone, use those. If you don't, you'll need to re-record.
The actual migration steps
1. Inventory your existing Play.ht projects. List every voice clone you're actively using, every project you need to maintain, and the scripts for each. Export audio files as your backup.
2. Re-upload your clone audio to ElevenLabs. Go to Voices > Add a New Voice > Instant Voice Clone. Upload at least one minute of clean, consistent audio, no music, no reverb, minimal background noise. Name it to match your Play.ht clone so you don't lose track.
3. Find replacements for standard voices. For Play.ht voices you use from the standard catalogue, spend time in ElevenLabs' Voice Library. Filter by gender, age, and use case. Audition with a sample from your actual script rather than the default preview text.
4. Calibrate your stability settings. The default ElevenLabs settings (Stability 0.5, Similarity 0.75) are a reasonable starting point. If your content is formal narration, push Stability to 0.70-0.80. If it's conversational or emotional content, lower Stability to 0.35-0.50 and listen to whether the expressiveness reads naturally.
5. Replace the WordPress integration. If you used Play.ht's plugin, set up an ElevenLabs API workflow. The ElevenLabs API is well-documented and there are Make/Zapier templates for common CMS workflows. Expect to spend a few hours on this, not a few minutes.
6. Test a full production piece. Generate a complete piece of content you'd normally produce in Play.ht, from script to finished audio file. Compare the output against your Play.ht reference. Adjust voice settings until the quality matches or exceeds what you had.
Gotchas you'll hit
The pricing model is different in a way that can increase costs if you regenerate sections frequently. Play.ht typically charges per character with monthly caps. ElevenLabs also charges per character, but if you're iterating on a 3,000-word script and regenerating sections five times before you're happy, you're generating 15,000+ words of audio to produce one final piece. Plan your ElevenLabs character budget around your iteration habits, not just your final output volume.
Voice clone consistency can vary between sessions. ElevenLabs' IVC model can produce slightly different output from the same text on different days, because the underlying model is periodically updated. If you need identical output for ongoing series content, note the model version you used for your first episode.
Some Play.ht voices have SSML support for fine-grained control over pauses, emphasis, and rate changes within a sentence. ElevenLabs does not support SSML. If your scripts rely on SSML tags for timing, you'll need to re-edit them to use punctuation and natural sentence breaks instead.
There's no mobile app for ElevenLabs' web interface. Play.ht has a more accessible interface for quick generations from a tablet or phone. If you generate audio on the go, this is a real friction point.
When NOT to switch
Play.ht's WordPress plugin integration is genuinely hard to replicate in ElevenLabs without technical setup work. If you're a solo blogger who relies on auto-generated audio for every post, the friction of rebuilding that workflow in ElevenLabs is substantial.
Play.ht also has broader language support in some regional accents and dialects. If your audience is in a market where ElevenLabs' voice selection is thin, check the voice library carefully before committing to the switch.
For most content creators who are running into limitations with emotional range or voice clone quality, ElevenLabs delivers a clear step up from Play.ht. The workflow shift from document-centric to API-centric is the main adjustment, and it's worth making if voice quality is central to your product.