Connect your ElevenLabs API key to Phersa

Phersa uses your own ElevenLabs API key for Voice Swap. ElevenLabs Voice Changer (speech to speech) turns the voice in a generated clip into your persona's ElevenLabs voice, and Phersa adds no markup.

Get your ElevenLabs API key

elevenlabs.io/app/developers/api-keys Open

Prices

ElevenLabs models in Phersa and what they cost

ElevenLabs models in Phersa with the price for each setting
Model	Setting	Price
ElevenLabs Voice Changer (Multilingual)	8 s clip	$0.016 ($0.002/s × 8s)
ElevenLabs Voice Changer (English)	8 s clip	$0.016 ($0.002/s × 8s)

Voice Swap is priced per second of the clip's audio, shown here for an 8-second clip.

Setup

How to connect an ElevenLabs API key to Phersa

phersa.com/elevenlabs

Step 1: Get an ElevenLabs API key

Sign in at elevenlabs.io, open API keys and create a key. It starts with sk_. Voice Swap uses the credits on your ElevenLabs plan, and using your own cloned voice or a Voice Library voice through the API needs a paid plan (Starter or higher).
elevenlabs.io/app/developers/api-keys →
Step 2: Add the key in Phersa

In Phersa, open Settings, go to API keys and choose ElevenLabs. Paste the key and click Save & verify.
Step 3: Let Phersa check it

Phersa first checks that the key starts with sk_, then tests it with ElevenLabs. A key ElevenLabs rejects is not stored. Saved keys are encrypted, and once connected the ElevenLabs panel in Settings shows the characters left on your plan.
Step 4: Give your persona a voice

In ElevenLabs, copy the voice ID of a voice in your library. In Phersa, open Personas and the persona, find Voice under Identity, paste the ID and click Add.
Step 5: Swap the voice

In the app, open a clip with dialogue that already has a generated video and choose Voice Swap. Pick the model and whether to keep background sound; Phersa shows the cost before you swap, such as $0.016 for an 8-second clip.

MCP server details

Provider keys are added in Phersa Settings only. The Phersa MCP server cannot read, add or change them.
settings-keys-status shows whether ElevenLabs is connected.
Voice Swap needs a persona with an ElevenLabs voice. Use personas-voices-list to find one and personas-write to save it to the persona.
Voice Swap runs through jobs-create with type voice, which starts as a dry run. Ask the user before swapping for real, because it spends the characters on their ElevenLabs plan.
Full MCP guide: phersa.com/mcp

On ElevenLabs

How Voice Swap works on ElevenLabs

Phersa splits the clip's audio into the voice and the background sound, sends the voice through ElevenLabs Voice Changer, then puts the new voice back on the original video, so the lipsync timing stays. If the split is not possible, the whole track is converted.
Background sound: Keep mixes the clip's own sound back under the new voice (the default); Remove ships the voice on its own.
Two models: Multilingual v2 (recommended, the default) and English v2, at the same price.
Voices come from your own ElevenLabs library: paste the ID of a premade, cloned, designed or shared voice. Phersa does not clone voices. Cloning starts on the ElevenLabs Starter plan, and Voice Library voices are not available through the API on the Free plan.
ElevenLabs bills Voice Changer at 1,000 characters per minute of audio. Phersa shows the Pro plan rate, $0.12 per minute ($0.002 per second of the clip); your plan's rate can differ.
ElevenLabs is the only Voice Swap provider in Phersa. Clips without dialogue have no Voice Swap, and images and video (Nano Banana 2, Gemini Omni 1.1 Flash, GPT Image 2.5 Sunburst, Grok Imagine Image 2.0 and Grok Imagine Video 1.5) run on other providers such as Agent Platform and Google AI Studio.

ElevenLabs questions

What is ElevenLabs?

ElevenLabs is an AI research and product company that builds voice models for speech, transcription and voice changing, including the speech-to-speech Voice Changer. It was founded in 2022 by Mati Staniszewski and Piotr Dabkowski and sells monthly plans with a shared credit allowance.

Visit elevenlabs.io ↗

How much does Voice Swap cost with ElevenLabs?

Phersa shows $0.002 per second of the clip's audio: $0.016 for an 8-second clip and $0.12 per minute. That is ElevenLabs' Pro plan rate for Voice Changer, which bills 1,000 characters per minute of audio; your plan's rate can differ.

Why use ElevenLabs’ Voice Changer?

To keep your AI character’s voice consistent. A video model makes the voice for each clip on its own, so the same AI persona can sound different from one clip to the next. Voice Changer replaces the voice in every clip with the one ElevenLabs voice you picked, and keeps the timing and delivery of the original line, so the lip sync still matches. You can keep the clip’s background sound or remove it.

Does Phersa charge extra on top of ElevenLabs?

No. Voice Swap uses the characters on your own ElevenLabs plan, and Phersa adds no markup. The Phersa plan is a separate $99 per month.

See Phersa pricing →

Does ElevenLabs have a free tier or free credits?

Yes, with limits. The Free plan includes 10,000 credits a month but no commercial license, Voice Library voices are not available through the API on Free, and voice cloning starts on Starter. For Voice Swap with your own voice, plan on Starter or higher. Terms can change.

Check ElevenLabs pricing ↗

Do I need ElevenLabs for Voice Swap in Phersa?

Yes. Voice Swap runs only on ElevenLabs. Image and video generation use other providers, such as Kie, fal or Google AI Studio.

See Kie for images and video →

How do I connect my ElevenLabs API key to Phersa?

Create a key at elevenlabs.io/app/developers/api-keys. In Phersa, open Settings, go to API keys, choose ElevenLabs and paste it. Then paste a voice ID into your persona under Identity, Voice.

See the setup steps →

Does Phersa clone my voice?

No. You paste the ID of a voice that is already in your ElevenLabs library, including voices you cloned or designed in ElevenLabs, and Phersa uses that voice for your persona. Cloning a voice in ElevenLabs needs the Starter plan or higher.

Should I use Multilingual v2 or English v2?

Multilingual v2 is the default and the recommended model. English v2 is for English-only dialogue. Both cost the same.

Does Voice Swap keep the background sound?

By default, yes. Phersa separates the voice from the background, swaps the voice and mixes the background back in. Choose Remove to get the new voice on its own.

What balance does Phersa show for ElevenLabs?

The characters left on your ElevenLabs plan: your plan limit minus what you have used, shown in the ElevenLabs panel in Settings. Voice Swap uses about 1,000 characters per minute of audio.

Is my ElevenLabs API key safe in Phersa?

Phersa tests the key with ElevenLabs before saving it, stores it encrypted and uses it only for Voice Swap, checking voice IDs and showing your character balance. AI agents connected over MCP cannot read or change it. Removing the key in Settings deletes it permanently.