# How to move a project off a retiring ElevenLabs default voice

Updated 2026-11-06

Move a project off a retiring ElevenLabs Default voice by inventorying its voice IDs, auditioning replacements against your real script, and switching one complete deliverable at a time. ElevenLabs says Default voices expire on December 31, 2026 and are available only to accounts created before March 2026. I'd start with Titan for a serious explainer or Heisenberg for a long lesson, depending on the role the retiring voice had.

Those are choices from my own catalog. An ElevenLabs replacement and a different voice that suits your script are two separate options. Hear both before you decide. The [Default voices hub](https://voicefieldguide.com/elevenlabs-default-voices) helps with the casting side; this page covers the work that keeps a migration from breaking a project.

## What needs an inventory before the voice changes?

Make a list of every place that can generate speech, including the ones nobody has touched in months. A video editor's project may point at finished audio, while a scheduled automation still calls the old ID. Updating the editor will leave the automation unchanged.

For each use, record the voice ID, model ID, script location, output format and destination. Add the person or process that approves the result. Keep this file beside the project without putting API keys in it.

Search your source code, workflow exports and project settings for the actual ID. Searching the display name alone misses scripts that store only the identifier. If an automation lets you set a voice globally and override it in individual steps, inspect both locations. A hidden override is exactly the sort of thing that passes a quick smoke test.

Finished audio deserves its own column. Mark it as either approved and archived, or expected to be regenerated. You need to know which recordings can remain in an existing edit and which workflows depend on access to the voice next month.

## Should I use the official replacement or choose a new voice?

First check the current ElevenLabs notice for that voice. An older help article describes automatic routing for some Legacy IDs. That history does not promise automatic routing for every Default voice retiring now. I would not build a December release around that assumption.

My shortlist starts with the job the voice must do.

### Titan for weight without a theatrical performance

[Titan](https://voicefieldguide.com/elevenlabs-voices/titan-deep-bold-and-powerful) fits serious explainers and documentary copy. It measured about 170 words per minute in my clips. Plan from that, then time a take of your own. Its consonants give names and figures somewhere to land. The cost is gravity. A breezy tutorial can start sounding like a warning, so include one lighter paragraph in the audition if your series uses humor.

### Heisenberg when the listener needs time

[Heisenberg](https://voicefieldguide.com/elevenlabs-voices/heisenberg-polished-and-clear) is my calmer candidate for educational narration. It measured about 175 words per minute in my clips, which may still differ from the old narrator, so a scene cut to that voice's timing may need rewriting. Don't stretch the entire recording until it fits and assume the problem is solved. Compare the breathing room around the explanation.

### Dexter for something the listener must do

[Dexter](https://voicefieldguide.com/elevenlabs-voices/dexter-customer-support-pro) suits support and onboarding, reading at about 190 words per minute in my clips before deliberate action gaps. It gives instructions evenly. Hype is its weak spot. If the old narrator also read the product trailer, let that trailer have a separate audition.

If Titan fits your project, [add Titan to My Voices](https://elevenlabs.io/app/voice-lab/share/e0464333b21941625fcdd28b2ce050bc74911f6bd69cec0d6861a8bd9e24032b/dtSEyYGNJqjrtBArPCVZ) before testing generation in your account.

## What should the replacement audition contain?

Use a passage that exposes failure. A polished opening with five easy words proves very little. Include the hardest proper noun, a figure spoken in full, and a sentence near the end of a section. Add a line whose timing must match a visible action.

For a fictional project, this is a useful test passage.

> The Northbridge archive moved to its new building in nineteen eighty-six. Open the catalog and choose the collection name. Wait for the list to appear before selecting a record. Some records contain a photograph, a map, and a separate written description. Listen to the last sentence carefully. It should finish as a complete thought, with enough space afterward for the viewer to read the date on screen.

Replace that passage with equivalent material from your project. Keep the script identical across the audition candidates. Otherwise a shorter rewrite can make a slower voice seem faster and disguise what changed.

## How do I switch without losing the approved version?

Keep the old audio and its edit available. Render the candidate into a new folder, with a filename that identifies the voice and revision. Change the ID in one test workflow before changing the shared production setting.

Play the finished export, including music and captions. Check the first sentence after every edit. A replacement can have acceptable total duration while leaving one name clipped at a cut or one caption well ahead of the spoken line.

Then change the production setting and run the same job through the real output destination. Confirm the file that arrives there, not just the preview in the generator. Keep the old configuration so you can diagnose a failed delivery, even when the retiring voice will eventually stop being a usable fallback.

## What counts as a finished migration?

I would close it when every generating workflow uses an intentional current ID, the approved replacements are archived, and the owner of each deliverable has heard its final export. Record unresolved pronunciations explicitly. A successful HTTP request tells you that audio was generated; it doesn't tell you that your client's company name survived the change.

Schedule the final rehearsal before the retirement date. Leave time for the awkward paragraph. There is always an awkward paragraph.

## Audition details

These picks compare my fifteen voices. I own them, and ElevenLabs may pay me when you use them.

### Titan: Serious explainer replacement

A low read with distinct consonants for a project that needs weight. Light entertainment can sound too grave.

- Voice ID: `dtSEyYGNJqjrtBArPCVZ`
- Voice page: https://voicefieldguide.com/elevenlabs-voices/titan-deep-bold-and-powerful
- Add to My Voices: https://elevenlabs.io/app/voice-lab/share/e0464333b21941625fcdd28b2ce050bc74911f6bd69cec0d6861a8bd9e24032b/dtSEyYGNJqjrtBArPCVZ
- This guide's Eleven v4 clip: https://voicefieldguide.com/audio/v4/guide--elevenlabs-default-voice-migration--titan-deep-bold-and-powerful.mp3

Sample line: Before we change the narrator, save the original passage. The replacement needs to carry the same dates and names, and leave enough room for the edit we already have.

### Heisenberg: Long educational reads

An even option for lessons and chapters, provided the existing timeline has room for the pauses after each line.

- Voice ID: `iEBOK9alpKauGRvBSsFi`
- Voice page: https://voicefieldguide.com/elevenlabs-voices/heisenberg-polished-and-clear
- Add to My Voices: https://elevenlabs.io/app/voice-lab/share/e0464333b21941625fcdd28b2ce050bc74911f6bd69cec0d6861a8bd9e24032b/iEBOK9alpKauGRvBSsFi
- This guide's Eleven v4 clip: https://voicefieldguide.com/audio/v4/guide--elevenlabs-default-voice-migration--heisenberg-polished-and-clear.mp3

Sample line: This is the paragraph to audition first. It contains the difficult name, the longest explanation, and the quiet ending that has to sound like it belongs to the same speaker.

### Dexter: Support and onboarding

Even instruction pacing helps a replacement preserve usable gaps between actions. It won't supply a sales pitch's excitement.

- Voice ID: `Smxkoz0xiOoHo5WcSskf`
- Voice page: https://voicefieldguide.com/elevenlabs-voices/dexter-customer-support-pro
- Add to My Voices: https://elevenlabs.io/app/voice-lab/share/2c79c9f85b01099b7eb10d1340f7f6d43d132db92fe0ede23cd10eef2678f8f1/Smxkoz0xiOoHo5WcSskf
- This guide's Eleven v4 clip: https://voicefieldguide.com/audio/v4/guide--elevenlabs-default-voice-migration--dexter-customer-support-pro.mp3

Sample line: Open the saved project and play the original instructions once. Then compare the new recording. Check that each action finishes before the next instruction begins.

## Sources

- [ElevenLabs: Accessing Default voices](https://elevenlabs.io/docs/help-center/product/voices/my-voices/how-do-i-access-eleven-labs-default-voices)
- [ElevenLabs: How voices are updated](https://help.elevenlabs.io/hc/en-us/articles/20939463028881-How-are-voices-updated-changes)
- [ElevenLabs: Create speech API](https://elevenlabs.io/docs/api-reference/text-to-speech/convert)

Source page: https://voicefieldguide.com/guides/elevenlabs-default-voice-migration
